{ "schema": "ontology_article_evidence_matrix/v1", "generated_at": "2026-06-25T04:35:00+08:00", "kb_metrics": { "source_count": 803, "core_sources": 150, "candidate_sources": 332, "low_priority_sources": 321, "selected_sources": 382, "chunks": 2157, "claims": 1307, "entities": 1804, "relations": 16721, "agent_report_chunks": 438 }, "sections": [ { "id": "ontology_foundations", "title": "Ontology foundations: explicit commitments, not taxonomies", "thesis": "A serious ontology is an explicit and reusable commitment about what exists in a domain, how things are identified, what relations can hold, and what constraints make a model coherent.", "source_ids": [ "core-gruber-1993-portable-ontology", "oa-https-doi-org-10-1006-ijhc-1995-1081", "core-guarino-1998-formal-ontology-information-systems", "guarino_oberle_staab_2009_what_is_ontology", "core-noy-mcguinness-2001-ontology-101", "core-baader-2003-description-logic-handbook", "core-bfo-iso-21838-2" ], "claims": [ "Ontology should be defined as an explicit specification of a conceptualization, not as a synonym for any taxonomy or data catalog.", "Formal ontology adds discipline to information systems by clarifying intended meanings and category commitments behind symbols.", "Ontology engineering requires scope, reuse, competency questions, constraints, evaluation, and maintenance.", "Upper ontologies and description logics matter because representation choices determine what can be inferred or checked." ], "caveats": [ "Do not imply every enterprise ontology is a complete formal ontology.", "Distinguish ontology, taxonomy, controlled vocabulary, schema, and graph database." ], "retrieval_concepts": [ "ontology", "formal_ontology", "knowledge_representation", "ontology_engineering", "validation" ], "sources": [ { "source_id": "core-gruber-1993-portable-ontology", "title": "A Translation Approach to Portable Ontology Specifications", "bucket": "academic", "quality_signal": "peer_reviewed_seminal", "year": 1993, "url": "https://doi.org/10.1006/knac.1993.1008", "triage_tier": "core", "triage_score": 98, "sample_claims": [ "Ontologies enable knowledge sharing by making conceptual commitments explicit.", "A portable ontology specification can support translation across representation systems.", "Ontologies support sharing and reuse of knowledge among agents and systems.", "An ontology can be treated as an explicit specification of a conceptualization." ], "sample_chunks": [ "chunk-core-gruber-1993-portable-ontology-001", "chunk-core-gruber-1993-portable-ontology-002", "chunk-core-gruber-1993-portable-ontology-003", "chunk-core-gruber-1993-portable-ontology-004" ] }, { "source_id": "oa-https-doi-org-10-1006-ijhc-1995-1081", "title": "Toward principles for the design of ontologies used for knowledge sharing?", "bucket": "academic", "quality_signal": "peer_reviewed_seminal", "year": 1995, "url": "https://doi.org/10.1006/ijhc.1995.1081", "triage_tier": "core", "triage_score": 114, "sample_claims": [ "Ontology design should balance clarity, coherence, extendibility, encoding neutrality, and minimal commitment.", "Explicit design criteria make ontologies easier to share and reuse.", "Ontology design should make intended meanings clear and coherent.", "A reusable ontology should avoid unnecessary representation-specific commitments." ], "sample_chunks": [ "chunk-oa-https-doi-org-10-1006-ijhc-1995-1081-001", "chunk-oa-https-doi-org-10-1006-ijhc-1995-1081-002", "chunk-oa-https-doi-org-10-1006-ijhc-1995-1081-003", "chunk-oa-https-doi-org-10-1006-ijhc-1995-1081-004" ] }, { "source_id": "core-guarino-1998-formal-ontology-information-systems", "title": "Formal Ontology in Information Systems", "bucket": "academic", "quality_signal": "peer_reviewed_seminal", "year": 1998, "url": "https://dl.acm.org/doi/10.5555/521720.521722", "triage_tier": "core", "triage_score": 104, "sample_claims": [ "Formal ontology helps clarify conceptual distinctions in information systems.", "Ontology is useful when data structures need meaning-preserving integration." ], "sample_chunks": [ "chunk-core-guarino-1998-formal-ontology-information-systems-001", "chunk-core-guarino-1998-formal-ontology-information-systems-002", "chunk-core-guarino-1998-formal-ontology-information-systems-003" ] }, { "source_id": "guarino_oberle_staab_2009_what_is_ontology", "title": "What Is an Ontology?", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2009, "url": "https://doi.org/10.1007/978-3-540-92673-3_0", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "The word ontology has distinct philosophical and computational meanings.", "Computational ontologies formalize aspects of a conceptualization for a purpose.", "Not every taxonomy, vocabulary, or graph has the same level of ontological commitment." ], "sample_chunks": [ "chunk-guarino_oberle_staab_2009_what_is_ontology-001", "chunk-guarino_oberle_staab_2009_what_is_ontology-002", "chunk-guarino_oberle_staab_2009_what_is_ontology-003", "chunk-guarino_oberle_staab_2009_what_is_ontology-004" ] }, { "source_id": "core-noy-mcguinness-2001-ontology-101", "title": "Ontology Development 101: A Guide to Creating Your First Ontology", "bucket": "technical", "quality_signal": "widely_cited", "year": 2001, "url": "https://protege.stanford.edu/publications/ontology_development/ontology101.pdf", "triage_tier": "core", "triage_score": 98, "sample_claims": [ "Ontology development is iterative and should begin with scope and competency questions.", "Classes, relations, properties, constraints, and instances should be modeled deliberately.", "Ontology construction should begin with domain and scope questions.", "Existing ontologies should be considered before building from scratch." ], "sample_chunks": [ "chunk-core-noy-mcguinness-2001-ontology-101-001", "chunk-core-noy-mcguinness-2001-ontology-101-002", "chunk-core-noy-mcguinness-2001-ontology-101-003", "chunk-core-noy-mcguinness-2001-ontology-101-004" ] }, { "source_id": "core-baader-2003-description-logic-handbook", "title": "The Description Logic Handbook: Theory, Implementation, and Applications", "bucket": "books", "quality_signal": "scholarly_book", "year": 2003, "url": "https://www.cambridge.org/core/books/description-logic-handbook", "triage_tier": "core", "triage_score": 106, "sample_claims": [ "Description logics balance expressivity and decidable reasoning.", "Ontology languages can be grounded in formal semantics and automated inference." ], "sample_chunks": [ "chunk-core-baader-2003-description-logic-handbook-001", "chunk-core-baader-2003-description-logic-handbook-002", "chunk-core-baader-2003-description-logic-handbook-003" ] }, { "source_id": "core-bfo-iso-21838-2", "title": "Information technology - Top-level ontologies - Part 2: Basic Formal Ontology (BFO)", "bucket": "technical", "quality_signal": "official_standard", "year": 2021, "url": "https://www.iso.org/standard/74572.html", "triage_tier": "core", "triage_score": 108, "sample_claims": [ "BFO supplies upper-level categories for domain ontology construction.", "A top-level ontology can improve cross-domain consistency and interoperability.", "BFO is standardized as a top-level ontology for broad interoperability.", "Top-level ontology standards can support consistent domain ontology development." ], "sample_chunks": [ "chunk-core-bfo-iso-21838-2-001", "chunk-core-bfo-iso-21838-2-002", "chunk-core-bfo-iso-21838-2-003", "chunk-core-bfo-iso-21838-2-004" ] } ] }, { "id": "standards_stack", "title": "Standards stack: from graph data to validation and governance", "thesis": "The standards ecosystem gives ontology-backed AI systems a neutral vocabulary for identity, graph representation, querying, reasoning, validation, provenance, cataloging, and policy.", "source_ids": [ "core-w3c-rdf-11-concepts", "core-w3c-owl2-overview", "core-w3c-sparql-11-query", "w3c_shacl_2017", "w3c_skos_reference_2009", "phase2-w3c-prov-o-2013", "p2-llm-ont-032", "phase14-schemaorg-evolution-structured-data-web-2016", "phase14-linkml-open-data-modeling-framework-2026", "phase14-sssom-ontological-mappings-2022", "phase4-std-w3c-r2rml-2012", "phase15-w3c-direct-mapping-relational-data-rdf-2012", "phase15-dimou-2014-rml-integrated-rdf-mappings", "phase15-rml-specification-heterogeneous-rdf-mapping-2024", "phase15-yarrrml-human-readable-rdf-generation-rules-2025", "phase4-std-w3c-odrl-model-2018", "phase4-std-w3c-dqv-2016", "p2-llm-ont-036", "phase14-fair-data-point-metadata-publication-2023", "phase14-fair-digital-object-conceptual-model-2023", "phase16-w3c-sparql11-federated-query-2013", "phase6-odcs-310-open-data-contract-standard", "phase6-odps-41-open-data-product-specification", "phase6-openlineage-docs-specification", "phase6-omg-dmn-15", "phase4-std-dpv-2-0-2024", "phase16-oasis-xacml-3-policy-access-control-2013" ], "claims": [ "RDF supplies a graph data model; OWL supplies formal ontology semantics; SPARQL supplies graph querying.", "SHACL is a practical bridge from semantic models to production conformance checks.", "SKOS is appropriate for lightweight concept schemes and controlled vocabularies.", "PROV-O, DCAT, DQV, ODRL, and DPROD extend semantic infrastructure into provenance, cataloging, quality, usage policy, and data product governance.", "LinkML, SSSOM, Schema.org, FAIR Data Point, and FAIR Digital Objects show how schemas, mappings, vocabularies, and metadata services make semantics operational beyond OWL/RDF alone.", "R2RML, Direct Mapping, RML, and YARRRML make the data-to-graph mapping layer explicit, reviewable, and reusable.", "Federated SPARQL query, privacy vocabularies, access-control policy, data contracts, lineage events, and decision-model standards make the semantic layer operational rather than merely descriptive." ], "caveats": [ "Standards show mechanisms, not automatic adoption or successful governance.", "OWL reasoning, SHACL validation, and policy enforcement are different layers and should not be collapsed." ], "retrieval_concepts": [ "semantic_web", "semantic_schema", "ontology_mapping_metadata", "mapping_language", "machine_actionable_metadata", "validation", "provenance", "data_quality", "usage_policy", "data_product" ], "sources": [ { "source_id": "core-w3c-rdf-11-concepts", "title": "RDF 1.1 Concepts and Abstract Syntax", "bucket": "technical", "quality_signal": "official_standard", "year": 2014, "url": "https://www.w3.org/TR/rdf11-concepts", "triage_tier": "core", "triage_score": 126, "sample_claims": [ "RDF represents information as graph triples.", "IRIs and literals provide a Web-scale naming and value model for semantic data.", "RDF represents information as graph-based triples.", "IRIs provide global identifiers for resources and properties." ], "sample_chunks": [ "chunk-core-w3c-rdf-11-concepts-001", "chunk-core-w3c-rdf-11-concepts-002", "chunk-core-w3c-rdf-11-concepts-003", "chunk-core-w3c-rdf-11-concepts-004" ] }, { "source_id": "core-w3c-owl2-overview", "title": "OWL 2 Web Ontology Language Document Overview", "bucket": "technical", "quality_signal": "official_standard", "year": 2012, "url": "https://www.w3.org/TR/owl2-overview", "triage_tier": "core", "triage_score": 108, "sample_claims": [ "OWL 2 supports formal ontology modeling with reasoning semantics.", "OWL profiles support different computational tradeoffs for applications.", "OWL 2 provides formal languages for expressing ontologies on the Web.", "OWL profiles trade expressivity for computational properties and implementation needs." ], "sample_chunks": [ "chunk-core-w3c-owl2-overview-001", "chunk-core-w3c-owl2-overview-002", "chunk-core-w3c-owl2-overview-003", "chunk-core-w3c-owl2-overview-004" ] }, { "source_id": "core-w3c-sparql-11-query", "title": "SPARQL 1.1 Query Language", "bucket": "technical", "quality_signal": "official_standard", "year": 2013, "url": "https://www.w3.org/TR/sparql11-query", "triage_tier": "core", "triage_score": 114, "sample_claims": [ "SPARQL provides standardized graph query over RDF datasets.", "Graph pattern matching supports retrieval across linked semantic data.", "SPARQL queries RDF graphs through graph patterns.", "Property paths and federation allow traversal and distributed querying." ], "sample_chunks": [ "chunk-core-w3c-sparql-11-query-001", "chunk-core-w3c-sparql-11-query-002", "chunk-core-w3c-sparql-11-query-003", "chunk-core-w3c-sparql-11-query-004" ] }, { "source_id": "w3c_shacl_2017", "title": "Shapes Constraint Language (SHACL)", "bucket": "technical", "quality_signal": "official_standard", "year": 2017, "url": "https://www.w3.org/TR/shacl", "triage_tier": "core", "triage_score": 126, "sample_claims": [ "SHACL separates data graphs from shapes graphs used for validation.", "Constraint validation can check cardinality, datatypes, classes, paths, and custom conditions.", "RDF graph quality can be tested before downstream use.", "SHACL validates RDF graph data against declared constraints." ], "sample_chunks": [ "chunk-w3c_shacl_2017-001", "chunk-w3c_shacl_2017-002", "chunk-w3c_shacl_2017-003", "chunk-w3c_shacl_2017-004" ] }, { "source_id": "w3c_skos_reference_2009", "title": "SKOS Simple Knowledge Organization System Reference", "bucket": "technical", "quality_signal": "official_standard", "year": 2009, "url": "https://www.w3.org/TR/skos-reference", "triage_tier": "core", "triage_score": 114, "sample_claims": [ "SKOS models concepts, labels, notation, semantic relations, and concept schemes.", "SKOS supports lightweight knowledge organization without full OWL commitment.", "SKOS can publish controlled vocabularies as linked data.", "SKOS makes controlled vocabularies and taxonomies publishable and linkable as Web data." ], "sample_chunks": [ "chunk-w3c_skos_reference_2009-001", "chunk-w3c_skos_reference_2009-002", "chunk-w3c_skos_reference_2009-003", "chunk-w3c_skos_reference_2009-004" ] }, { "source_id": "phase2-w3c-prov-o-2013", "title": "PROV-O: The PROV Ontology", "bucket": "technical", "quality_signal": "official_standard", "year": 2013, "url": "https://www.w3.org/TR/prov-o", "triage_tier": "core", "triage_score": 120, "sample_claims": [ "PROV-O provides a standard ontology for entities, activities, agents, and their provenance relations.", "Provenance is essential when knowledge graphs are generated, transformed, or used as evidence.", "AI knowledge bases need provenance if generated assertions must remain auditable." ], "sample_chunks": [ "chunk-phase2-w3c-prov-o-2013-001", "chunk-phase2-w3c-prov-o-2013-002", "chunk-phase2-w3c-prov-o-2013-003", "chunk-phase2-w3c-prov-o-2013-004" ] }, { "source_id": "p2-llm-ont-032", "title": "Data Catalog Vocabulary (DCAT) - Version 3", "bucket": "technical", "quality_signal": "official_standard", "year": 2024, "url": "https://www.w3.org/TR/vocab-dcat-3", "triage_tier": "candidate", "triage_score": 52, "sample_claims": [ "Standard catalog metadata improves discoverability and interoperability of datasets and data services.", "DCAT can act as a metadata layer for decentralized data publishing and federated discovery.", "Catalog semantics are a practical bridge between data management and ontology-backed applications." ], "sample_chunks": [ "chunk-p2-llm-ont-032-001", "chunk-p2-llm-ont-032-002", "chunk-p2-llm-ont-032-003", "chunk-p2-llm-ont-032-004" ] }, { "source_id": "phase14-schemaorg-evolution-structured-data-web-2016", "title": "Schema.org: Evolution of Structured Data on the Web", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2016, "url": "https://doi.org/10.1145/2857274.2857276", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Web-scale structured data requires common vocabularies that publishers and consumers can share.", "Schema.org demonstrates a pragmatic vocabulary ecosystem built for widespread publication and application use.", "Common schemas become more necessary as data volume and application diversity increase." ], "sample_chunks": [ "chunk-phase14-schemaorg-evolution-structured-data-web-2016-001", "chunk-phase14-schemaorg-evolution-structured-data-web-2016-002", "chunk-phase14-schemaorg-evolution-structured-data-web-2016-003", "chunk-phase14-schemaorg-evolution-structured-data-web-2016-004" ] }, { "source_id": "phase14-linkml-open-data-modeling-framework-2026", "title": "LinkML: an open data modeling framework", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2026, "url": "https://doi.org/10.1093/gigascience/giaf152", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "LinkML provides an approachable modeling framework for authoring, validating, and sharing structured data schemas.", "The framework connects software-engineering schema needs with linked-data and ontology-ready semantics.", "Typed schemas and validation help make scientific and enterprise data more reusable, interoperable, and less ambiguous for downstream computation." ], "sample_chunks": [ "chunk-phase14-linkml-open-data-modeling-framework-2026-001", "chunk-phase14-linkml-open-data-modeling-framework-2026-002", "chunk-phase14-linkml-open-data-modeling-framework-2026-003", "chunk-phase14-linkml-open-data-modeling-framework-2026-004" ] }, { "source_id": "phase14-sssom-ontological-mappings-2022", "title": "A Simple Standard for Sharing Ontological Mappings (SSSOM)", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2022, "url": "https://doi.org/10.1093/database/baac035", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Ontology mappings require machine-readable metadata that makes mapping semantics, provenance, and uncertainty explicit.", "SSSOM models mappings as pairwise assertions grouped into mapping sets with standard slots.", "FAIR exchange of mappings is a prerequisite for reliable ontology integration, reconciliation, and cross-resource data reuse." ], "sample_chunks": [ "chunk-phase14-sssom-ontological-mappings-2022-001", "chunk-phase14-sssom-ontological-mappings-2022-002", "chunk-phase14-sssom-ontological-mappings-2022-003", "chunk-phase14-sssom-ontological-mappings-2022-004" ] }, { "source_id": "phase4-std-w3c-r2rml-2012", "title": "R2RML: RDB to RDF Mapping Language", "bucket": "technical", "quality_signal": "official_standard", "year": 2012, "url": "https://www.w3.org/TR/r2rml", "triage_tier": "core", "triage_score": 76, "sample_claims": [ "R2RML standardizes relational-to-RDF mappings for customized graph generation.", "Mapping rules make the connection between operational tables and semantic graph entities explicit.", "R2RML is a key standards bridge between existing databases and ontology-backed knowledge graphs." ], "sample_chunks": [ "chunk-phase4-std-w3c-r2rml-2012-001", "chunk-phase4-std-w3c-r2rml-2012-002", "chunk-phase4-std-w3c-r2rml-2012-003", "chunk-phase4-std-w3c-r2rml-2012-004" ] }, { "source_id": "phase15-w3c-direct-mapping-relational-data-rdf-2012", "title": "A Direct Mapping of Relational Data to RDF", "bucket": "technical", "quality_signal": "official_standard", "year": 2012, "url": "https://www.w3.org/TR/rdb-direct-mapping", "triage_tier": "core", "triage_score": 76, "sample_claims": [ "Direct Mapping defines a default RDF representation for relational databases.", "The recommendation makes table, row, primary-key, foreign-key, and literal translation rules explicit.", "Default mappings provide a baseline before customized R2RML-style semantic mappings are designed." ], "sample_chunks": [ "chunk-phase15-w3c-direct-mapping-relational-data-rdf-2012-001", "chunk-phase15-w3c-direct-mapping-relational-data-rdf-2012-002", "chunk-phase15-w3c-direct-mapping-relational-data-rdf-2012-003", "chunk-phase15-w3c-direct-mapping-relational-data-rdf-2012-004" ] }, { "source_id": "phase15-dimou-2014-rml-integrated-rdf-mappings", "title": "RML: A Generic Language for Integrated RDF Mappings of Heterogeneous Data", "bucket": "academic", "quality_signal": "peer_reviewed_survey", "year": 2014, "url": "https://ceur-ws.org/Vol-1184/ldow2014_paper_01.pdf", "triage_tier": "core", "triage_score": 70, "sample_claims": [ "RML generalizes R2RML to heterogeneous data sources and serializations.", "Declarative mappings can make heterogeneous data integration into RDF more reusable and less tool-specific.", "RML exposes mapping logic that would otherwise remain hidden in extraction code." ], "sample_chunks": [ "chunk-phase15-dimou-2014-rml-integrated-rdf-mappings-001", "chunk-phase15-dimou-2014-rml-integrated-rdf-mappings-002", "chunk-phase15-dimou-2014-rml-integrated-rdf-mappings-003", "chunk-phase15-dimou-2014-rml-integrated-rdf-mappings-004" ] }, { "source_id": "phase15-rml-specification-heterogeneous-rdf-mapping-2024", "title": "RDF Mapping Language (RML)", "bucket": "technical", "quality_signal": "official_docs", "year": 2024, "url": "https://rml.io/specs/rml", "triage_tier": "core", "triage_score": 72, "sample_claims": [ "RML expresses mapping rules from heterogeneous data structures to RDF.", "The language extends R2RML while broadening source support beyond relational databases.", "Source-agnostic mapping specifications improve interoperability and repeatability of KG construction pipelines." ], "sample_chunks": [ "chunk-phase15-rml-specification-heterogeneous-rdf-mapping-2024-001", "chunk-phase15-rml-specification-heterogeneous-rdf-mapping-2024-002", "chunk-phase15-rml-specification-heterogeneous-rdf-mapping-2024-003", "chunk-phase15-rml-specification-heterogeneous-rdf-mapping-2024-004" ] }, { "source_id": "phase15-yarrrml-human-readable-rdf-generation-rules-2025", "title": "YARRRML: Human Readable Text-Based Representation for Declarative Generation Rules", "bucket": "technical", "quality_signal": "official_docs", "year": 2025, "url": "https://w3id.org/yarrrml/spec", "triage_tier": "core", "triage_score": 72, "sample_claims": [ "YARRRML provides a human-readable representation for declarative RDF generation rules.", "The format can simplify authoring mappings over multiple existing data sources.", "Usable mapping syntax helps make semantic integration maintainable by data teams, not only semantic-web experts." ], "sample_chunks": [ "chunk-phase15-yarrrml-human-readable-rdf-generation-rules-2025-001", "chunk-phase15-yarrrml-human-readable-rdf-generation-rules-2025-002", "chunk-phase15-yarrrml-human-readable-rdf-generation-rules-2025-003", "chunk-phase15-yarrrml-human-readable-rdf-generation-rules-2025-004" ] }, { "source_id": "phase4-std-w3c-odrl-model-2018", "title": "ODRL Information Model 2.2", "bucket": "technical", "quality_signal": "official_standard", "year": 2018, "url": "https://www.w3.org/TR/odrl-model", "triage_tier": "low_priority", "triage_score": 25, "sample_claims": [], "sample_chunks": [] }, { "source_id": "phase4-std-w3c-dqv-2016", "title": "Data on the Web Best Practices: Data Quality Vocabulary", "bucket": "technical", "quality_signal": "official_standard", "year": 2016, "url": "https://www.w3.org/TR/vocab-dqv", "triage_tier": "candidate", "triage_score": 52, "sample_claims": [], "sample_chunks": [ "chunk-phase4-std-w3c-dqv-2016-001" ] }, { "source_id": "p2-llm-ont-036", "title": "Data Products Ontology Specification Version 1.0 beta", "bucket": "technical", "quality_signal": "official_docs", "year": 2025, "url": "https://www.omg.org/spec/DPROD/1.0/Beta1/About-DPROD", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [], "sample_chunks": [] }, { "source_id": "phase14-fair-data-point-metadata-publication-2023", "title": "FAIR Data Point: A FAIR-Oriented Approach for Metadata Publication", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2023, "url": "https://doi.org/10.1162/dint_a_00160", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "FAIR metadata publication requires semantically rich, machine-actionable metadata services.", "FAIR Data Point defines architecture and metadata provision patterns for discovering and searching metadata about digital objects.", "The approach connects linked data, FAIR principles, metadata schemas, and service-level discovery." ], "sample_chunks": [ "chunk-phase14-fair-data-point-metadata-publication-2023-001", "chunk-phase14-fair-data-point-metadata-publication-2023-002", "chunk-phase14-fair-data-point-metadata-publication-2023-003", "chunk-phase14-fair-data-point-metadata-publication-2023-004" ] }, { "source_id": "phase14-fair-digital-object-conceptual-model-2023", "title": "Towards a Conceptual Model for the FAIR Digital Object Framework", "bucket": "academic", "quality_signal": "preprint", "year": 2023, "url": "https://arxiv.org/abs/2302.11894", "triage_tier": "low_priority", "triage_score": 36, "sample_claims": [ "FAIR Digital Objects aim to make digital objects identifiable, describable, and explorable according to FAIR principles.", "An ontology-driven conceptual model is needed to clarify expected FDO entities, relations, and behaviours.", "FDOs extend the machine-actionable metadata argument from datasets to broader digital-object infrastructure." ], "sample_chunks": [ "chunk-phase14-fair-digital-object-conceptual-model-2023-001", "chunk-phase14-fair-digital-object-conceptual-model-2023-002", "chunk-phase14-fair-digital-object-conceptual-model-2023-003", "chunk-phase14-fair-digital-object-conceptual-model-2023-004" ] }, { "source_id": "phase16-w3c-sparql11-federated-query-2013", "title": "SPARQL 1.1 Federated Query", "bucket": "technical", "quality_signal": "official_standard", "year": 2013, "url": "https://www.w3.org/TR/sparql11-federated-query", "triage_tier": "core", "triage_score": 76, "sample_claims": [ "SPARQL federated query allows portions of a query to be delegated to remote SPARQL services.", "Federated query supports distributed semantic architectures where data remains under separate endpoints.", "Cross-organization ontology-backed AI can use federated query patterns as one retrieval mechanism." ], "sample_chunks": [ "chunk-phase16-w3c-sparql11-federated-query-2013-001", "chunk-phase16-w3c-sparql11-federated-query-2013-002", "chunk-phase16-w3c-sparql11-federated-query-2013-003", "chunk-phase16-w3c-sparql11-federated-query-2013-004" ] }, { "source_id": "phase6-odcs-310-open-data-contract-standard", "title": "Open Data Contract Standard v3.1.0", "bucket": "technical", "quality_signal": "official_docs", "year": 2026, "url": "https://bitol-io.github.io/open-data-contract-standard/v3.1.0", "triage_tier": "candidate", "triage_score": 66, "sample_claims": [ "Data contracts encode producer-consumer expectations in versionable YAML.", "Contracts combine structural schema with quality and operational commitments.", "The specification is designed for automation and validation." ], "sample_chunks": [ "chunk-phase6-odcs-310-open-data-contract-standard-001", "chunk-phase6-odcs-310-open-data-contract-standard-002", "chunk-phase6-odcs-310-open-data-contract-standard-003", "chunk-phase6-odcs-310-open-data-contract-standard-004" ] }, { "source_id": "phase6-odps-41-open-data-product-specification", "title": "Open Data Product Specification 4.1", "bucket": "technical", "quality_signal": "official_docs", "year": 2026, "url": "https://opendataproducts.org/v4.1", "triage_tier": "candidate", "triage_score": 54, "sample_claims": [ "Data products need a shared, machine-readable structure.", "ODPS includes operational and business metadata beyond dataset schema.", "Version 4.1 adds reference mechanisms for modular governance." ], "sample_chunks": [ "chunk-phase6-odps-41-open-data-product-specification-001", "chunk-phase6-odps-41-open-data-product-specification-002", "chunk-phase6-odps-41-open-data-product-specification-003", "chunk-phase6-odps-41-open-data-product-specification-004" ] }, { "source_id": "phase6-openlineage-docs-specification", "title": "OpenLineage Documentation and Specification", "bucket": "technical", "quality_signal": "official_docs", "year": 2026, "url": "https://openlineage.io/docs", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [ "Lineage should be emitted through shared event metadata rather than custom integrations.", "The core model identifies jobs, runs, and datasets consistently.", "Facets extend the model for system-specific metadata." ], "sample_chunks": [ "chunk-phase6-openlineage-docs-specification-001", "chunk-phase6-openlineage-docs-specification-002", "chunk-phase6-openlineage-docs-specification-003", "chunk-phase6-openlineage-docs-specification-004" ] }, { "source_id": "phase6-omg-dmn-15", "title": "Decision Model and Notation Specification Version 1.5", "bucket": "technical", "quality_signal": "official_standard", "year": 2024, "url": "https://www.omg.org/spec/DMN/1.5/About-DMN", "triage_tier": "candidate", "triage_score": 52, "sample_claims": [ "DMN bridges business decision design and technical implementation.", "Decision models can be shared between business analysts, developers, and decision owners.", "DMN is designed to complement BPMN process modeling." ], "sample_chunks": [ "chunk-phase6-omg-dmn-15-001", "chunk-phase6-omg-dmn-15-002", "chunk-phase6-omg-dmn-15-003", "chunk-phase6-omg-dmn-15-004" ] }, { "source_id": "phase4-std-dpv-2-0-2024", "title": "Data Privacy Vocabulary (DPV) Version 2.0", "bucket": "technical", "quality_signal": "official_docs", "year": 2024, "url": "https://www.w3.org/community/reports/dpvcg/CG-FINAL-dpv-20240801", "triage_tier": "core", "triage_score": 72, "sample_claims": [ "Privacy and data processing metadata can be expressed with shared machine-readable vocabularies.", "DPV gives taxonomies for purposes, processing operations, personal data categories, legal bases, rights, risks, and controls.", "AI context and retrieval layers can use privacy vocabularies to constrain use and disclosure." ], "sample_chunks": [ "chunk-phase4-std-dpv-2-0-2024-001", "chunk-phase4-std-dpv-2-0-2024-002", "chunk-phase4-std-dpv-2-0-2024-003", "chunk-phase4-std-dpv-2-0-2024-004" ] }, { "source_id": "phase16-oasis-xacml-3-policy-access-control-2013", "title": "eXtensible Access Control Markup Language (XACML) Version 3.0", "bucket": "technical", "quality_signal": "official_standard", "year": 2013, "url": "https://docs.oasis-open.org/xacml/3.0/xacml-3.0-core-spec-os-en.html", "triage_tier": "core", "triage_score": 76, "sample_claims": [ "Access-control decisions can be represented as machine-readable policies and request/response evaluations.", "Policies can include obligations and advice that accompany decisions.", "Agentic AI governance needs policy enforcement layers outside the LLM when tools can read or change state." ], "sample_chunks": [ "chunk-phase16-oasis-xacml-3-policy-access-control-2013-001", "chunk-phase16-oasis-xacml-3-policy-access-control-2013-002", "chunk-phase16-oasis-xacml-3-policy-access-control-2013-003", "chunk-phase16-oasis-xacml-3-policy-access-control-2013-004" ] } ] }, { "id": "ontology_evaluation_lifecycle", "title": "Ontology evaluation and lifecycle governance: quality before action", "thesis": "Ontology-backed AI is only as reliable as the conceptual quality, reporting discipline, versioning, and maintenance workflow behind its semantic layer.", "source_ids": [ "fernandez_lopez_gomez_perez_juristo_1997_methontology", "guarino_welty_2002_ontoclean", "phase2-poveda-2014-oops", "suarez_figueroa_2015_neon", "phase16-noy-klein-2004-ontology-evolution-schema-evolution", "phase16-miro-guidelines-minimum-information-reporting-ontology-2017", "phase16-fairsharing-community-standards-repositories-policies-2019", "phase7-ontology-development-kit-toolkit-2022", "phase7-robot-automating-ontology-workflows-2019", "phase6-lippolis-2025-llm-assisting-ontology-evaluation", "phase6-kampars-2025-llm-collaborative-ontology-design" ], "claims": [ "Ontology quality is not reducible to graph size or syntactic validity; conceptual checks such as OntoClean and pitfall scanning catch different classes of modeling errors.", "Ontology evolution differs from database schema evolution because changes alter shared meaning, dependent mappings, applications, and governance commitments.", "METHONTOLOGY, NeOn, MIRO, ODK, and ROBOT show that specification, reuse, reporting, release engineering, validation, and maintenance are core ontology lifecycle activities.", "LLM-assisted ontology engineering raises the value of lifecycle controls because generation makes weak structures easier to produce at scale." ], "caveats": [ "Automated pitfall scanners and reporting checklists improve governance but do not replace domain expert review.", "Ontology versioning and migration are operational programs, not one-time modeling decisions." ], "retrieval_concepts": [ "ontology_quality", "ontology_evaluation", "ontology_evolution", "ontology_reporting", "ontology_governance", "validation" ], "sources": [ { "source_id": "fernandez_lopez_gomez_perez_juristo_1997_methontology", "title": "METHONTOLOGY: From Ontological Art Towards Ontological Engineering", "bucket": "academic", "quality_signal": "widely_cited", "year": 1997, "url": "https://aaai.org/papers/0005-ss97-06-005-methontology-from-ontological-art-towards-ontological-engineering", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [], "sample_chunks": [] }, { "source_id": "guarino_welty_2002_ontoclean", "title": "Evaluating Ontological Decisions with OntoClean", "bucket": "academic", "quality_signal": "widely_cited", "year": 2002, "url": "https://doi.org/10.1145/503124.503150", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [ "Taxonomic hierarchies can contain subtle ontological errors.", "Meta-properties can expose mistaken subclass relations.", "Identity, unity, dependence, and rigidity matter for high-quality category systems.", "Ontology evaluation must include conceptual discipline, not only syntax or graph completeness." ], "sample_chunks": [ "chunk-guarino_welty_2002_ontoclean-001", "chunk-guarino_welty_2002_ontoclean-002", "chunk-guarino_welty_2002_ontoclean-003", "chunk-guarino_welty_2002_ontoclean-004" ] }, { "source_id": "phase2-poveda-2014-oops", "title": "OOPS! (OntOlogy Pitfall Scanner!): An On-line Tool for Ontology Evaluation", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2014, "url": "https://doi.org/10.1007/s13740-014-0028-y", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Ontology quality can be partially evaluated by detecting recurring modeling pitfalls.", "Automated pitfall scanning complements expert review, reasoner checks, and competency-question tests.", "LLM-accelerated ontology engineering increases the need for scalable quality gates.", "Ontology quality can be partially evaluated by detecting common modeling pitfalls." ], "sample_chunks": [ "chunk-phase2-poveda-2014-oops-001", "chunk-phase2-poveda-2014-oops-002", "chunk-phase2-poveda-2014-oops-003", "chunk-phase2-poveda-2014-oops-004" ] }, { "source_id": "suarez_figueroa_2015_neon", "title": "A Scenario-Based Methodology for Ontology Development", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2015, "url": "https://doi.org/10.3233/AO-150145", "triage_tier": "low_priority", "triage_score": 36, "sample_claims": [ "Modern ontology projects often build networks rather than isolated ontologies.", "Ontology reuse and reengineering require scenario-specific methods.", "Collaborative ontology development needs explicit process support.", "Ontology projects often build ontology networks rather than isolated ontologies." ], "sample_chunks": [ "chunk-suarez_figueroa_2015_neon-001", "chunk-suarez_figueroa_2015_neon-002", "chunk-suarez_figueroa_2015_neon-003", "chunk-suarez_figueroa_2015_neon-004" ] }, { "source_id": "phase16-noy-klein-2004-ontology-evolution-schema-evolution", "title": "Ontology Evolution: Not the Same as Schema Evolution", "bucket": "academic", "quality_signal": "widely_cited", "year": 2004, "url": "https://www.sciencedirect.com/science/article/pii/S1570826804000224", "triage_tier": "candidate", "triage_score": 66, "sample_claims": [ "Ontology evolution is not the same as database schema evolution because ontology changes alter shared meanings and commitments.", "Versioning, change propagation, dependent artifacts, and collaborative negotiation are central to ontology maintenance.", "AI systems using ontology-backed memory or tools need explicit handling of semantic drift and change impact.", "DOI should be rechecked from the publisher page before formal submission." ], "sample_chunks": [ "chunk-phase16-noy-klein-2004-ontology-evolution-schema-evolution-001", "chunk-phase16-noy-klein-2004-ontology-evolution-schema-evolution-002", "chunk-phase16-noy-klein-2004-ontology-evolution-schema-evolution-003", "chunk-phase16-noy-klein-2004-ontology-evolution-schema-evolution-004" ] }, { "source_id": "phase16-miro-guidelines-minimum-information-reporting-ontology-2017", "title": "MIRO: Guidelines for Minimum Information for the Reporting of an Ontology", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2017, "url": "https://doi.org/10.1186/s13326-017-0172-7", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Ontology publications should report enough information for reuse, evaluation, and comparison.", "Scope, intended use, knowledge acquisition, reuse, evaluation, and maintenance are part of responsible ontology reporting.", "Transparent ontology reporting supports scientific reuse and downstream computational trust." ], "sample_chunks": [ "chunk-phase16-miro-guidelines-minimum-information-reporting-ontology-2017-001", "chunk-phase16-miro-guidelines-minimum-information-reporting-ontology-2017-002", "chunk-phase16-miro-guidelines-minimum-information-reporting-ontology-2017-003", "chunk-phase16-miro-guidelines-minimum-information-reporting-ontology-2017-004" ] }, { "source_id": "phase16-fairsharing-community-standards-repositories-policies-2019", "title": "FAIRsharing as a Community Approach to Standards, Repositories and Policies", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2019, "url": "https://doi.org/10.1038/s41587-019-0080-8", "triage_tier": "candidate", "triage_score": 54, "sample_claims": [ "Scientific data reuse depends on discoverable standards, repositories, and policies.", "Registries help communities converge on shared reporting and metadata resources.", "Agents that reason over scientific data need machine-discoverable standards and repository context." ], "sample_chunks": [ "chunk-phase16-fairsharing-community-standards-repositories-policies-2019-001", "chunk-phase16-fairsharing-community-standards-repositories-policies-2019-002", "chunk-phase16-fairsharing-community-standards-repositories-policies-2019-003", "chunk-phase16-fairsharing-community-standards-repositories-policies-2019-004" ] }, { "source_id": "phase7-ontology-development-kit-toolkit-2022", "title": "Ontology Development Kit: a toolkit for building, maintaining and standardizing biomedical ontologies", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2022, "url": "https://doi.org/10.1093/database/baac087", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Ontology development requires repeatable workflows for release, testing, imports, and documentation.", "ODK packages community best practices into reusable project templates.", "Automated ontology engineering tooling makes ontology maintenance more scalable and auditable." ], "sample_chunks": [ "chunk-phase7-ontology-development-kit-toolkit-2022-001", "chunk-phase7-ontology-development-kit-toolkit-2022-002", "chunk-phase7-ontology-development-kit-toolkit-2022-003", "chunk-phase7-ontology-development-kit-toolkit-2022-004" ] }, { "source_id": "phase7-robot-automating-ontology-workflows-2019", "title": "ROBOT: A Tool for Automating Ontology Workflows", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2019, "url": "https://doi.org/10.1186/s12859-019-3002-3", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Ontology workflows can be automated through repeatable command-line operations.", "ROBOT supports reasoning, merging, extraction, validation reporting, and release workflows.", "Automated reports help make ontology quality issues visible and actionable." ], "sample_chunks": [ "chunk-phase7-robot-automating-ontology-workflows-2019-001", "chunk-phase7-robot-automating-ontology-workflows-2019-002", "chunk-phase7-robot-automating-ontology-workflows-2019-003", "chunk-phase7-robot-automating-ontology-workflows-2019-004" ] }, { "source_id": "phase6-lippolis-2025-llm-assisting-ontology-evaluation", "title": "Large Language Models Assisting Ontology Evaluation", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2025, "url": "https://doi.org/10.1007/978-3-032-09527-5_27", "triage_tier": "candidate", "triage_score": 54, "sample_claims": [ "LLMs can assist competency-question verification at roughly human-user performance levels in the reported setting.", "Hybrid expert-plus-LLM evaluation can reduce ontology evaluation effort while retaining expert control.", "The paper contributes a dataset of competency questions paired with ontologies and ontology stories." ], "sample_chunks": [ "chunk-phase6-lippolis-2025-llm-assisting-ontology-evaluation-001", "chunk-phase6-lippolis-2025-llm-assisting-ontology-evaluation-002", "chunk-phase6-lippolis-2025-llm-assisting-ontology-evaluation-003", "chunk-phase6-lippolis-2025-llm-assisting-ontology-evaluation-004" ] }, { "source_id": "phase6-kampars-2025-llm-collaborative-ontology-design", "title": "LLM-Supported Collaborative Ontology Design for Data and Knowledge Management Platforms", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2025, "url": "https://doi.org/10.3389/fdata.2025.1676477", "triage_tier": "low_priority", "triage_score": 36, "sample_claims": [], "sample_chunks": [] } ] }, { "id": "scientific_ontology_infrastructure", "title": "Scientific ontology infrastructure: community knowledge as machine substrate", "thesis": "In biology and data-intensive science, ontologies already function as maintained infrastructure for identifiers, annotations, causal models, phenotype descriptions, workflow metadata, and ontology-aware machine learning.", "source_ids": [ "core-gene-ontology-2000", "phase7-gene-ontology-knowledgebase-2023", "phase7-go-cam-causal-activity-modeling-2019", "smith_etal_2007_obo_foundry", "phase7-obo-foundry-2021-operationalizing-open-data-principles", "phase7-ontology-development-kit-toolkit-2022", "phase7-robot-automating-ontology-workflows-2019", "noy_etal_2009_bioportal", "phase7-ols4-ontology-lookup-service-2025", "phase7-ontobee-linked-ontology-server-2016", "phase7-bioregistry-biomedical-entity-identification-2022", "wilkinson_etal_2016_fair", "phase7-ro-crate-packaging-research-artefacts-2022", "phase7-human-phenotype-ontology-2024", "phase7-monarch-initiative-2024", "phase7-edam-bioinformatics-ontology-2013", "phase7-deepgo-se-semantic-entailment-2024" ], "claims": [ "The Gene Ontology shows how a community-maintained ontology can become durable scientific infrastructure for annotation, reuse, and computation.", "GO-CAM, HPO, Monarch, Cell Ontology, Uberon, EDAM, BioPortal, OLS4, Ontobee, and Bioregistry show that scientific ontology is an ecosystem of identifiers, services, mappings, workflows, and curation practices.", "OBO Foundry, ODK, and ROBOT make ontology governance and maintenance operational through principles, templates, reports, releases, and automated checks.", "DeepGO, DeepGOPlus, and DeepGO-SE show ontology can define the output space, constraints, and semantics of neural prediction, not only label finished results." ], "caveats": [ "Biomedical ontology success depends on years of community curation; it should not be presented as something LLMs can cheaply reproduce.", "Scientific ontology infrastructure proves the value of shared semantics, but not that every enterprise ontology has the same openness or accountability." ], "retrieval_concepts": [ "scientific_ontology_infrastructure", "gene_ontology", "biomedical_ontology", "ontology_workflow", "ontology_identifier_governance", "curated_knowledge_infrastructure", "ontology_aware_ai", "ai_for_science" ], "sources": [ { "source_id": "core-gene-ontology-2000", "title": "Gene Ontology: Tool for the Unification of Biology", "bucket": "academic", "quality_signal": "peer_reviewed_seminal", "year": 2000, "url": "https://doi.org/10.1038/75556", "triage_tier": "core", "triage_score": 98, "sample_claims": [ "A shared ontology enables consistent annotation of gene products across databases and species.", "Community-maintained ontologies can become scientific infrastructure.", "Shared biological vocabularies enable cross-database annotation and comparison.", "Ontology can unify data across species and research communities." ], "sample_chunks": [ "chunk-core-gene-ontology-2000-001", "chunk-core-gene-ontology-2000-002", "chunk-core-gene-ontology-2000-003", "chunk-core-gene-ontology-2000-004" ] }, { "source_id": "phase7-gene-ontology-knowledgebase-2023", "title": "The Gene Ontology knowledgebase in 2023", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2023, "url": "https://doi.org/10.1093/genetics/iyad031", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "The Gene Ontology knowledgebase is a large computable resource for gene and gene-product function.", "GO-CAM extends term annotations into structured causal activity models.", "Schema validation and archival releases are necessary for reproducibility and traceability.", "Community ontology maintenance is an ongoing scientific infrastructure process." ], "sample_chunks": [ "chunk-phase7-gene-ontology-knowledgebase-2023-001", "chunk-phase7-gene-ontology-knowledgebase-2023-002", "chunk-phase7-gene-ontology-knowledgebase-2023-003", "chunk-phase7-gene-ontology-knowledgebase-2023-004" ] }, { "source_id": "phase7-go-cam-causal-activity-modeling-2019", "title": "Gene Ontology Causal Activity Modeling (GO-CAM) moves beyond GO annotations to structured descriptions of biological functions and systems", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2019, "url": "https://doi.org/10.1038/s41588-019-0500-1", "triage_tier": "candidate", "triage_score": 54, "sample_claims": [ "GO-CAM represents biological function as structured causal activity models.", "GO-CAM links gene products, molecular activities, biological processes, and causal relations.", "Structured causal annotations make computational reasoning over biological systems more precise." ], "sample_chunks": [ "chunk-phase7-go-cam-causal-activity-modeling-2019-001", "chunk-phase7-go-cam-causal-activity-modeling-2019-002", "chunk-phase7-go-cam-causal-activity-modeling-2019-003", "chunk-phase7-go-cam-causal-activity-modeling-2019-004" ] }, { "source_id": "smith_etal_2007_obo_foundry", "title": "The OBO Foundry: Coordinated Evolution of Ontologies to Support Biomedical Data Integration", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2007, "url": "https://doi.org/10.1038/nbt1346", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Ontology interoperability requires governance principles, not only shared technology.", "Orthogonality reduces overlap and inconsistency among ontology modules.", "Open, versioned, well-defined ontologies can support biomedical data integration." ], "sample_chunks": [ "chunk-smith_etal_2007_obo_foundry-001", "chunk-smith_etal_2007_obo_foundry-002", "chunk-smith_etal_2007_obo_foundry-003", "chunk-smith_etal_2007_obo_foundry-004" ] }, { "source_id": "phase7-obo-foundry-2021-operationalizing-open-data-principles", "title": "OBO Foundry in 2021: Operationalizing Open Data Principles to Evaluate Ontologies", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2021, "url": "https://doi.org/10.1093/database/baab069", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Ontology ecosystems need measurable governance principles.", "Open biomedical ontologies are evaluated across identifiers, definitions, relations, versioning, openness, and maintenance.", "Interoperability depends on governance and lifecycle discipline, not only representation language." ], "sample_chunks": [ "chunk-phase7-obo-foundry-2021-operationalizing-open-data-principles-001", "chunk-phase7-obo-foundry-2021-operationalizing-open-data-principles-002", "chunk-phase7-obo-foundry-2021-operationalizing-open-data-principles-003", "chunk-phase7-obo-foundry-2021-operationalizing-open-data-principles-004" ] }, { "source_id": "phase7-ontology-development-kit-toolkit-2022", "title": "Ontology Development Kit: a toolkit for building, maintaining and standardizing biomedical ontologies", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2022, "url": "https://doi.org/10.1093/database/baac087", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Ontology development requires repeatable workflows for release, testing, imports, and documentation.", "ODK packages community best practices into reusable project templates.", "Automated ontology engineering tooling makes ontology maintenance more scalable and auditable." ], "sample_chunks": [ "chunk-phase7-ontology-development-kit-toolkit-2022-001", "chunk-phase7-ontology-development-kit-toolkit-2022-002", "chunk-phase7-ontology-development-kit-toolkit-2022-003", "chunk-phase7-ontology-development-kit-toolkit-2022-004" ] }, { "source_id": "phase7-robot-automating-ontology-workflows-2019", "title": "ROBOT: A Tool for Automating Ontology Workflows", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2019, "url": "https://doi.org/10.1186/s12859-019-3002-3", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Ontology workflows can be automated through repeatable command-line operations.", "ROBOT supports reasoning, merging, extraction, validation reporting, and release workflows.", "Automated reports help make ontology quality issues visible and actionable." ], "sample_chunks": [ "chunk-phase7-robot-automating-ontology-workflows-2019-001", "chunk-phase7-robot-automating-ontology-workflows-2019-002", "chunk-phase7-robot-automating-ontology-workflows-2019-003", "chunk-phase7-robot-automating-ontology-workflows-2019-004" ] }, { "source_id": "noy_etal_2009_bioportal", "title": "BioPortal: Ontologies and Integrated Data Resources at the Click of a Mouse", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2009, "url": "https://doi.org/10.1093/nar/gkp440", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [], "sample_chunks": [] }, { "source_id": "phase7-ols4-ontology-lookup-service-2025", "title": "OLS4: a new Ontology Lookup Service for a growing interdisciplinary knowledge ecosystem", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2025, "url": "https://doi.org/10.1093/bioinformatics/btaf279", "triage_tier": "candidate", "triage_score": 42, "sample_claims": [ "Ontology services need scalable search, metadata, term lookup, and API access.", "A growing interdisciplinary ontology ecosystem requires infrastructure beyond individual ontology files.", "Programmatic ontology access supports annotation, integration, and AI pipelines." ], "sample_chunks": [ "chunk-phase7-ols4-ontology-lookup-service-2025-001", "chunk-phase7-ols4-ontology-lookup-service-2025-002", "chunk-phase7-ols4-ontology-lookup-service-2025-003", "chunk-phase7-ols4-ontology-lookup-service-2025-004" ] }, { "source_id": "phase7-ontobee-linked-ontology-server-2016", "title": "Ontobee: A linked ontology data server to support ontology term dereferencing, linkage, query and integration", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2016, "url": "https://doi.org/10.1093/nar/gkw918", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [ "Ontology terms need dereferenceable web identifiers and linked-data access.", "Ontology servers can support SPARQL query, linkage, and integration across ontology resources.", "Linked ontology infrastructure improves reuse and machine access." ], "sample_chunks": [ "chunk-phase7-ontobee-linked-ontology-server-2016-001", "chunk-phase7-ontobee-linked-ontology-server-2016-002", "chunk-phase7-ontobee-linked-ontology-server-2016-003", "chunk-phase7-ontobee-linked-ontology-server-2016-004" ] }, { "source_id": "phase7-bioregistry-biomedical-entity-identification-2022", "title": "Unifying the identification of biomedical entities with the Bioregistry", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2022, "url": "https://doi.org/10.1038/s41597-022-01807-3", "triage_tier": "candidate", "triage_score": 42, "sample_claims": [ "Biomedical data integration requires consistent identifier prefixes and URI patterns.", "Registry mappings reduce ambiguity in entity references across resources.", "Identifier governance is a foundation for machine-actionable knowledge graphs." ], "sample_chunks": [ "chunk-phase7-bioregistry-biomedical-entity-identification-2022-001", "chunk-phase7-bioregistry-biomedical-entity-identification-2022-002", "chunk-phase7-bioregistry-biomedical-entity-identification-2022-003", "chunk-phase7-bioregistry-biomedical-entity-identification-2022-004" ] }, { "source_id": "wilkinson_etal_2016_fair", "title": "The FAIR Guiding Principles for Scientific Data Management and Stewardship", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2016, "url": "https://doi.org/10.1038/sdata.2016.18", "triage_tier": "candidate", "triage_score": 54, "sample_claims": [ "Machine-actionable metadata is central to reusable scientific data.", "Interoperability requires formal, shared, and broadly applicable languages for knowledge representation.", "Reusable data needs rich metadata, provenance, and clear usage conditions." ], "sample_chunks": [ "chunk-wilkinson_etal_2016_fair-001", "chunk-wilkinson_etal_2016_fair-002", "chunk-wilkinson_etal_2016_fair-003", "chunk-wilkinson_etal_2016_fair-004" ] }, { "source_id": "phase7-ro-crate-packaging-research-artefacts-2022", "title": "Packaging research artefacts with RO-Crate", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2022, "url": "https://doi.org/10.3233/ds-210053", "triage_tier": "candidate", "triage_score": 42, "sample_claims": [ "Research artifacts need portable machine-readable metadata packages.", "RO-Crate uses linked-data conventions to describe datasets, software, workflows, people, and provenance.", "Packaging metadata with artifacts improves reuse and automated processing." ], "sample_chunks": [ "chunk-phase7-ro-crate-packaging-research-artefacts-2022-001", "chunk-phase7-ro-crate-packaging-research-artefacts-2022-002", "chunk-phase7-ro-crate-packaging-research-artefacts-2022-003", "chunk-phase7-ro-crate-packaging-research-artefacts-2022-004" ] }, { "source_id": "phase7-human-phenotype-ontology-2024", "title": "The Human Phenotype Ontology in 2024: phenotypes around the world", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2024, "url": "https://doi.org/10.1093/nar/gkad1005", "triage_tier": "candidate", "triage_score": 42, "sample_claims": [ "HPO provides structured phenotype terms for clinical and translational research.", "Phenotype ontologies enable cross-resource computation over patients, diseases, genes, and variants.", "Global adoption requires translations, mappings, tooling, and community curation." ], "sample_chunks": [ "chunk-phase7-human-phenotype-ontology-2024-001", "chunk-phase7-human-phenotype-ontology-2024-002", "chunk-phase7-human-phenotype-ontology-2024-003", "chunk-phase7-human-phenotype-ontology-2024-004" ] }, { "source_id": "phase7-monarch-initiative-2024", "title": "The Monarch Initiative in 2024: an analytic platform integrating phenotypes, genes and diseases across species", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2024, "url": "https://doi.org/10.1093/nar/gkad1082", "triage_tier": "candidate", "triage_score": 42, "sample_claims": [ "Ontology mappings enable integrated analysis of phenotypes, genes, diseases, and model organism data.", "Cross-species phenotype and disease integration supports translational discovery.", "Graph-based biomedical platforms require shared identifiers, ontologies, and curation workflows." ], "sample_chunks": [ "chunk-phase7-monarch-initiative-2024-001", "chunk-phase7-monarch-initiative-2024-002", "chunk-phase7-monarch-initiative-2024-003", "chunk-phase7-monarch-initiative-2024-004" ] }, { "source_id": "phase7-edam-bioinformatics-ontology-2013", "title": "EDAM: an ontology of bioinformatics operations, types of data and identifiers, topics and formats", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2013, "url": "https://doi.org/10.1093/bioinformatics/btt113", "triage_tier": "candidate", "triage_score": 54, "sample_claims": [ "Bioinformatics tools and workflows benefit from shared ontology terms for operations, data, identifiers, topics, and formats.", "Tool annotation becomes more interoperable when inputs and outputs are semantically typed.", "Ontology supports workflow discovery and composition." ], "sample_chunks": [ "chunk-phase7-edam-bioinformatics-ontology-2013-001", "chunk-phase7-edam-bioinformatics-ontology-2013-002", "chunk-phase7-edam-bioinformatics-ontology-2013-003", "chunk-phase7-edam-bioinformatics-ontology-2013-004" ] }, { "source_id": "phase7-deepgo-se-semantic-entailment-2024", "title": "Protein function prediction as approximate semantic entailment", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2024, "url": "https://doi.org/10.1038/s42256-024-00795-w", "triage_tier": "low_priority", "triage_score": 36, "sample_claims": [ "Protein function prediction can be formulated as approximate semantic entailment over ontology axioms.", "Ontology semantics can improve both prediction structure and interpretability.", "Formal ontology can act as an explicit constraint and explanation layer for neural prediction." ], "sample_chunks": [ "chunk-phase7-deepgo-se-semantic-entailment-2024-001", "chunk-phase7-deepgo-se-semantic-entailment-2024-002", "chunk-phase7-deepgo-se-semantic-entailment-2024-003", "chunk-phase7-deepgo-se-semantic-entailment-2024-004" ] } ] }, { "id": "llm_kg_rag", "title": "LLM + KG + RAG: external semantic memory for generative systems", "thesis": "LLMs made language interfaces cheap, but reliable domain AI still needs external semantic memory, graph retrieval, validation, and provenance.", "source_ids": [ "arxiv-2306-08302v3", "oa-https-doi-org-10-1145-3447772", "oa-https-doi-org-10-1109-tnnls-2021-3070843", "core-lewis-2020-rag", "core-edge-2024-graphrag", "phase6-sharma-2025-og-rag", "phase6-oarga-2026-scientific-kg-ontology-open-llms", "phase6-zhou-2025-kg-rag-incompleteness", "phase6-xiang-2025-when-to-use-graphs-rag", "phase11-ali-2026-ontology-grounded-kg-clinical-hallucinations", "phase11-wu-2026-memgraphrag-memory-multi-agent-graphrag", "ont-ai-024", "ont-ai-022", "ont-ai-023", "phase2-gutierrez-2025-hipporag2", "phase5-liu-2025-ontology-guided-reverse-thinking-kgqa", "phase6-sui-2025-fidelis-kgqa" ], "claims": [ "LLM+KG integration is a recognized research agenda: KGs can ground LLMs, while LLMs can construct, complete, query, and align KGs.", "GraphRAG-style systems add entities, relations, paths, communities, and summaries above raw chunks.", "Ontology-grounded RAG is emerging as a specific branch where domain ontologies shape hypergraph or graph retrieval context.", "2026 work now connects ontology-grounded KGs to clinical hallucination mitigation and memory-based multi-agent GraphRAG to higher-quality graph construction.", "Professional-domain generation requires schemas, temporal relations, expert rules, provenance, and validation, not only nearest-neighbor passages.", "Ontology-guided KGQA shows ontology can guide reasoning paths, not just serve as a storage format.", "KG-RAG benchmarks now test when graph structure helps and how incompleteness degrades reasoning." ], "caveats": [ "Generated graphs are only as good as extraction, entity resolution, summarization, and provenance.", "Preprints such as KAG, LightRAG, and HippoRAG variants are architecture evidence; avoid treating them as settled field consensus." ], "retrieval_concepts": [ "llm_kg", "knowledge_graph", "rag", "ontology_grounded_rag", "hallucination_mitigation", "graphrag", "memgraphrag", "kgqa" ], "sources": [ { "source_id": "arxiv-2306-08302v3", "title": "Unifying Large Language Models and Knowledge Graphs: A Roadmap", "bucket": "academic", "quality_signal": "peer_reviewed_survey", "year": 2023, "url": "https://arxiv.org/abs/2306.08302", "triage_tier": "core", "triage_score": 120, "sample_claims": [ "KGs can improve LLM factuality, reasoning, interpretability, and domain adaptation.", "LLMs can help construct, complete, query, and reason over KGs.", "LLMs can support KG construction, completion, alignment, and question answering.", "Future systems require tight integration of parametric and symbolic knowledge." ], "sample_chunks": [ "chunk-arxiv-2306-08302v3-001", "chunk-arxiv-2306-08302v3-002", "chunk-arxiv-2306-08302v3-003", "chunk-arxiv-2306-08302v3-004" ] }, { "source_id": "oa-https-doi-org-10-1145-3447772", "title": "Knowledge Graphs", "bucket": "academic", "quality_signal": "peer_reviewed_survey", "year": 2021, "url": "https://dl.acm.org/doi/10.1145/3447772", "triage_tier": "core", "triage_score": 127, "sample_claims": [ "Knowledge graphs combine graph-structured data with semantics, identifiers, and sometimes formal constraints.", "KG ecosystems include representation, querying, reasoning, construction, quality, and embeddings.", "Knowledge graphs combine graph-structured data with schema, identity, semantics, and sometimes formal reasoning.", "Graph embeddings and symbolic reasoning address different aspects of graph-based knowledge." ], "sample_chunks": [ "chunk-oa-https-doi-org-10-1145-3447772-001", "chunk-oa-https-doi-org-10-1145-3447772-002", "chunk-oa-https-doi-org-10-1145-3447772-003", "chunk-oa-https-doi-org-10-1145-3447772-004" ] }, { "source_id": "oa-https-doi-org-10-1109-tnnls-2021-3070843", "title": "A Survey on Knowledge Graphs: Representation, Acquisition, and Applications", "bucket": "academic", "quality_signal": "peer_reviewed_survey", "year": 2021, "url": "https://ieeexplore.ieee.org/document/9416312", "triage_tier": "core", "triage_score": 121, "sample_claims": [ "Knowledge graph research spans representation learning, knowledge acquisition, completion, fusion, and reasoning.", "KGs support search, recommendation, question answering, and explainability.", "KG research spans representation, acquisition, fusion, completion, and application.", "Embedding and neural methods help with incomplete KGs but do not replace symbolic semantics." ], "sample_chunks": [ "chunk-oa-https-doi-org-10-1109-tnnls-2021-3070843-001", "chunk-oa-https-doi-org-10-1109-tnnls-2021-3070843-002", "chunk-oa-https-doi-org-10-1109-tnnls-2021-3070843-003", "chunk-oa-https-doi-org-10-1109-tnnls-2021-3070843-004" ] }, { "source_id": "core-lewis-2020-rag", "title": "Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks", "bucket": "academic", "quality_signal": "peer_reviewed_seminal", "year": 2020, "url": "https://arxiv.org/abs/2005.11401", "triage_tier": "core", "triage_score": 110, "sample_claims": [ "Retrieval can ground generation in external knowledge sources.", "Hybrid parametric/non-parametric memory improves knowledge-intensive tasks.", "Retrieval can improve knowledge-intensive generation tasks.", "Separating parametric and non-parametric memory supports knowledge updates." ], "sample_chunks": [ "chunk-core-lewis-2020-rag-001", "chunk-core-lewis-2020-rag-002", "chunk-core-lewis-2020-rag-003", "chunk-core-lewis-2020-rag-004" ] }, { "source_id": "core-edge-2024-graphrag", "title": "From Local to Global: A Graph RAG Approach to Query-Focused Summarization", "bucket": "academic", "quality_signal": "primary_source", "year": 2024, "url": "https://arxiv.org/abs/2404.16130", "triage_tier": "core", "triage_score": 114, "sample_claims": [ "Graph-structured indexes can improve synthesis over large corpora beyond local chunk retrieval.", "Community summaries provide higher-level retrieval units for broad questions.", "Graph structure can improve query-focused summarization over large corpora.", "Community detection and summaries support global questions not handled well by local chunk retrieval." ], "sample_chunks": [ "chunk-core-edge-2024-graphrag-001", "chunk-core-edge-2024-graphrag-002", "chunk-core-edge-2024-graphrag-003", "chunk-core-edge-2024-graphrag-004" ] }, { "source_id": "phase6-sharma-2025-og-rag", "title": "OG-RAG: Ontology-Grounded Retrieval-Augmented Generation for Large Language Models", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2025, "url": "https://aclanthology.org/2025.emnlp-main.1674", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Domain ontologies can improve retrieval context construction.", "The method reports gains in accurate fact recall, response correctness, attribution speed, and fact-based reasoning.", "Ontology-grounded RAG targets rule- and workflow-heavy domains where generic RAG underperforms." ], "sample_chunks": [ "chunk-phase6-sharma-2025-og-rag-001", "chunk-phase6-sharma-2025-og-rag-002", "chunk-phase6-sharma-2025-og-rag-003", "chunk-phase6-sharma-2025-og-rag-004" ] }, { "source_id": "phase6-oarga-2026-scientific-kg-ontology-open-llms", "title": "Scientific Knowledge Graph and Ontology Generation Using Open Large Language Models", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2026, "url": "https://pubs.rsc.org/en/content/articlelanding/2026/dd/d5dd00275c", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Generated ontologies improve GraphRAG answers over KG-only baselines in the reported experiment.", "Open LLMs can help structure scientific literature into knowledge graphs and ontology hierarchies.", "Ontology inclusion improves comprehensiveness, diversity, and empowerment criteria in the reported evaluation." ], "sample_chunks": [ "chunk-phase6-oarga-2026-scientific-kg-ontology-open-llms-001", "chunk-phase6-oarga-2026-scientific-kg-ontology-open-llms-002", "chunk-phase6-oarga-2026-scientific-kg-ontology-open-llms-003", "chunk-phase6-oarga-2026-scientific-kg-ontology-open-llms-004" ] }, { "source_id": "phase6-zhou-2025-kg-rag-incompleteness", "title": "Evaluating Knowledge Graph Based Retrieval Augmented Generation Methods under Knowledge Incompleteness", "bucket": "academic", "quality_signal": "preprint", "year": 2025, "url": "https://arxiv.org/abs/2504.05163", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [], "sample_chunks": [] }, { "source_id": "phase6-xiang-2025-when-to-use-graphs-rag", "title": "When to Use Graphs in RAG: A Comprehensive Analysis for Graph Retrieval-Augmented Generation", "bucket": "academic", "quality_signal": "preprint_benchmark", "year": 2025, "url": "https://arxiv.org/abs/2506.05690", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [ "GraphRAG does not always outperform vanilla RAG.", "Graph structure helps most under hierarchical retrieval and deeper contextual reasoning needs.", "Evaluation should cover graph construction, retrieval, and generation end to end." ], "sample_chunks": [ "chunk-phase6-xiang-2025-when-to-use-graphs-rag-001", "chunk-phase6-xiang-2025-when-to-use-graphs-rag-002", "chunk-phase6-xiang-2025-when-to-use-graphs-rag-003", "chunk-phase6-xiang-2025-when-to-use-graphs-rag-004" ] }, { "source_id": "phase11-ali-2026-ontology-grounded-kg-clinical-hallucinations", "title": "Ontology-grounded knowledge graphs for mitigating hallucinations in large language models for clinical question answering", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2026, "url": "https://doi.org/10.1016/j.jbi.2026.104993", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Ontology-grounded knowledge graphs are being evaluated as a mitigation strategy for hallucinations in clinical LLM question answering.", "Biomedical QA is a high-consequence setting where semantic grounding, provenance, and domain constraints matter more than open-ended language fluency.", "The article's reference network connects ontology-grounded QA to SPARQL generation, KG-guided RAG, COVID-19 knowledge graphs, and dense passage retrieval." ], "sample_chunks": [ "chunk-phase11-ali-2026-ontology-grounded-kg-clinical-hallucinations-001", "chunk-phase11-ali-2026-ontology-grounded-kg-clinical-hallucinations-002", "chunk-phase11-ali-2026-ontology-grounded-kg-clinical-hallucinations-003", "chunk-phase11-ali-2026-ontology-grounded-kg-clinical-hallucinations-004" ] }, { "source_id": "phase11-wu-2026-memgraphrag-memory-multi-agent-graphrag", "title": "MemGraphRAG: Memory-based Multi-Agent System for Graph Retrieval-Augmented Generation", "bucket": "academic", "quality_signal": "preprint", "year": 2026, "url": "https://arxiv.org/abs/2606.00610", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [ "GraphRAG can capture structural relationships for complex reasoning, but fragment-level graph extraction can create inconsistent or disconnected graphs.", "Shared memory across agents gives graph construction a global corpus context.", "Memory-aware hierarchical retrieval is proposed to exploit the constructed graph for retrieval-augmented generation.", "The paper reports improvements over baseline models while keeping efficiency comparable." ], "sample_chunks": [ "chunk-phase11-wu-2026-memgraphrag-memory-multi-agent-graphrag-001", "chunk-phase11-wu-2026-memgraphrag-memory-multi-agent-graphrag-002", "chunk-phase11-wu-2026-memgraphrag-memory-multi-agent-graphrag-003", "chunk-phase11-wu-2026-memgraphrag-memory-multi-agent-graphrag-004" ] }, { "source_id": "ont-ai-024", "title": "KAG: Boosting LLMs in Professional Domains via Knowledge Augmented Generation", "bucket": "academic", "quality_signal": "secondary_source", "year": 2024, "url": "https://arxiv.org/abs/2409.13731", "triage_tier": "core", "triage_score": 96, "sample_claims": [ "Professional-domain LLM systems need structured knowledge and reasoning beyond generic RAG.", "Knowledge organization and retrieval strategy materially affect answer reliability.", "Domain schemas and knowledge bases can improve controllability.", "Professional-domain RAG needs logic, temporal relations, numerical constraints, and expert rules beyond vector similarity." ], "sample_chunks": [ "chunk-ont-ai-024-001", "chunk-ont-ai-024-002", "chunk-ont-ai-024-003", "chunk-ont-ai-024-004" ] }, { "source_id": "ont-ai-022", "title": "LightRAG: Simple and Fast Retrieval-Augmented Generation", "bucket": "academic", "quality_signal": "secondary_source", "year": 2024, "url": "https://arxiv.org/abs/2410.05779", "triage_tier": "core", "triage_score": 102, "sample_claims": [ "Graph-enhanced retrieval can be made more efficient than heavier graph pipelines.", "Combining local and global context can improve answer quality.", "RAG systems must balance retrieval depth, latency, and graph maintenance cost.", "Flat vector retrieval misses interdependencies that graph structures can expose." ], "sample_chunks": [ "chunk-ont-ai-022-001", "chunk-ont-ai-022-002", "chunk-ont-ai-022-003", "chunk-ont-ai-022-004" ] }, { "source_id": "ont-ai-023", "title": "HippoRAG: Neurobiologically Inspired Long-Term Memory for Large Language Models", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2024, "url": "https://arxiv.org/abs/2405.14831", "triage_tier": "core", "triage_score": 104, "sample_claims": [ "LLM memory benefits from associative structures beyond flat vector search.", "Graph-like retrieval supports multi-hop and cross-document reasoning.", "Memory quality depends on extraction, indexing, and retrieval mechanisms.", "Knowledge graphs can act as associative memory structures for LLM retrieval." ], "sample_chunks": [ "chunk-ont-ai-023-001", "chunk-ont-ai-023-002", "chunk-ont-ai-023-003", "chunk-ont-ai-023-004" ] }, { "source_id": "phase2-gutierrez-2025-hipporag2", "title": "From RAG to Memory: Non-Parametric Continual Learning for Large Language Models", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2025, "url": "https://arxiv.org/abs/2502.14802", "triage_tier": "core", "triage_score": 110, "sample_claims": [ "RAG can be reframed as non-parametric continual learning rather than only answer retrieval.", "Graph-augmented memory must preserve factual accuracy while improving associativity and sense-making.", "Continual AI memory requires organization and retrieval policies, not only embedding stores.", "RAG can be reframed as non-parametric memory rather than transient document lookup." ], "sample_chunks": [ "chunk-phase2-gutierrez-2025-hipporag2-001", "chunk-phase2-gutierrez-2025-hipporag2-002", "chunk-phase2-gutierrez-2025-hipporag2-003", "chunk-phase2-gutierrez-2025-hipporag2-004" ] }, { "source_id": "phase5-liu-2025-ontology-guided-reverse-thinking-kgqa", "title": "Ontology-Guided Reverse Thinking Makes Large Language Models Stronger on Knowledge Graph Question Answering", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2025, "url": "https://aclanthology.org/2025.acl-long.741", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Knowledge graph question answering still struggles with multi-hop reasoning when entity vector matching misses abstract question purposes.", "Ontology-Guided Reverse Thinking constructs reasoning paths from purposes back to conditions.", "The method uses LLMs to extract purpose and condition labels, constructs label reasoning paths based on KG ontology, and uses those paths to guide retrieval.", "Experiments on WebQSP and CWQ report state-of-the-art performance and improved LLM KGQA capability." ], "sample_chunks": [ "chunk-phase5-liu-2025-ontology-guided-reverse-thinking-kgqa-001", "chunk-phase5-liu-2025-ontology-guided-reverse-thinking-kgqa-002", "chunk-phase5-liu-2025-ontology-guided-reverse-thinking-kgqa-003", "chunk-phase5-liu-2025-ontology-guided-reverse-thinking-kgqa-004" ] }, { "source_id": "phase6-sui-2025-fidelis-kgqa", "title": "FiDeLiS: Faithful Reasoning in Large Language Models for Knowledge Graph Question Answering", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2025, "url": "https://aclanthology.org/2025.findings-acl.436", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Verifiable reasoning paths reduce hallucination in KGQA.", "Training-free path retrieval and beam search can improve accuracy and efficiency.", "LLMs can score reasoning paths as deductive reasoning tasks." ], "sample_chunks": [ "chunk-phase6-sui-2025-fidelis-kgqa-001", "chunk-phase6-sui-2025-fidelis-kgqa-002", "chunk-phase6-sui-2025-fidelis-kgqa-003", "chunk-phase6-sui-2025-fidelis-kgqa-004" ] } ] }, { "id": "ontology_engineering_llms", "title": "LLMs as ontology engineering assistants", "thesis": "LLMs can accelerate term extraction, relation typing, matching, schema-guided extraction, and draft ontology generation, but they increase the need for review, benchmarks, and constraints.", "source_ids": [ "phase2-li-garijo-poveda-2025-llm-oe-review", "ont-ai-004", "p2-llm-ont-006", "phase3-llm-ontmem-023", "phase2-llms4om-2024", "phase2-agent-om-2023", "p2-llm-ont-007", "p2-llm-ont-030", "phase6-lippolis-2025-llm-assisting-ontology-evaluation", "phase6-zhao-2025-llm-ontology-requirements-engineering", "phase6-lippolis-2025-domain-specific-ontology-generation-llms", "phase6-kampars-2025-llm-collaborative-ontology-design", "phase11-oyewale-soru-2026-ontoekg-enterprise-ontology-construction", "phase11-soares-wassermann-2026-specific-domain-ontology-construction-llms", "p2-llm-ont-006" ], "claims": [ "LLMs can support ontology learning, matching, population, relation extraction, and schema-guided knowledge extraction.", "Ontology engineering with LLMs should be workflow-based and human-in-the-loop rather than single-prompt generation.", "Evaluation must measure graph-level usability, consistency, alignment quality, and hallucination, not only text plausibility.", "Domain ontologies and schemas can constrain extraction, but they do not eliminate expert review.", "Recent peer-reviewed work is moving upstream and downstream of generation: requirements elicitation, competency-question verification, expert validation, and scholarly relation classification.", "2026 preprints sharpen the enterprise angle: OntoEKG decomposes ontology construction into extraction and entailment, while expert evaluation of LLM-built domain ontologies confirms that coherent drafts still require refinement." ], "caveats": [ "Many LLM ontology engineering studies are preprints or early benchmarks.", "High extraction scores do not guarantee maintainable ontology design." ], "retrieval_concepts": [ "ontology_learning", "ontology_matching", "ontology_engineering", "enterprise_ontology_construction", "ontology_evaluation", "schema_alignment", "llm" ], "sources": [ { "source_id": "phase2-li-garijo-poveda-2025-llm-oe-review", "title": "Large Language Models for Ontology Engineering: A Systematic Literature Review", "bucket": "academic", "quality_signal": "peer_reviewed_survey", "year": 2025, "url": "https://www.semantic-web-journal.net/content/large-language-models-ontology-engineering-systematic-literature-review", "triage_tier": "core", "triage_score": 108, "sample_claims": [ "LLMs are being applied across multiple ontology engineering tasks, but evaluation and reliability remain uneven.", "Ontology engineering with LLMs requires human oversight, task decomposition, and quality controls.", "The literature is shifting from isolated prompting experiments toward workflow-level OE assistance.", "LLMs are being used as ontology engineers, domain experts, and evaluators across the ontology lifecycle." ], "sample_chunks": [ "chunk-phase2-li-garijo-poveda-2025-llm-oe-review-001", "chunk-phase2-li-garijo-poveda-2025-llm-oe-review-002", "chunk-phase2-li-garijo-poveda-2025-llm-oe-review-003", "chunk-phase2-li-garijo-poveda-2025-llm-oe-review-004" ] }, { "source_id": "ont-ai-004", "title": "LLMs4OL: Large Language Models for Ontology Learning", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2023, "url": "https://arxiv.org/abs/2307.16648", "triage_tier": "core", "triage_score": 92, "sample_claims": [ "LLMs can assist ontology learning tasks by exploiting broad language knowledge.", "LLM outputs still require validation because generated ontology elements may be inconsistent or unsupported.", "Ontology learning benchmarks need adaptation for generative models.", "LLMs can be evaluated on concrete ontology learning subtasks rather than vague ontology generation claims." ], "sample_chunks": [ "chunk-ont-ai-004-001", "chunk-ont-ai-004-002", "chunk-ont-ai-004-003", "chunk-ont-ai-004-004" ] }, { "source_id": "p2-llm-ont-006", "title": "Large Language Models for Scholarly Ontology Generation: An Extensive Analysis in the Engineering Field", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2025, "url": "https://doi.org/10.1016/j.ipm.2025.104262", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [ "LLMs can classify broader, narrower, same-as, and other topic relations with strong performance under some prompting strategies.", "Smaller or quantized models may be competitive when prompts are tuned for the task.", "Scholarly ontologies need relation-specific evaluation rather than generic text similarity scores.", "LLMs can support relationship classification among scholarly concepts." ], "sample_chunks": [ "chunk-p2-llm-ont-006-001", "chunk-p2-llm-ont-006-002", "chunk-p2-llm-ont-006-003", "chunk-p2-llm-ont-006-004" ] }, { "source_id": "phase3-llm-ontmem-023", "title": "LLM-empowered Knowledge Graph Construction: A Survey", "bucket": "academic", "quality_signal": "peer_reviewed_survey", "year": 2025, "url": "https://arxiv.org/abs/2510.20345", "triage_tier": "candidate", "triage_score": 64, "sample_claims": [ "LLM-empowered KGC requires attention to both extraction quality and downstream graph usability.", "Ontology and schema guidance can improve consistency, but hallucination and entity resolution remain core risks.", "Evaluation of KGC pipelines should measure graph-level quality, not just extraction micro-metrics." ], "sample_chunks": [ "chunk-phase3-llm-ontmem-023-001", "chunk-phase3-llm-ontmem-023-002", "chunk-phase3-llm-ontmem-023-003", "chunk-phase3-llm-ontmem-023-004" ] }, { "source_id": "phase2-llms4om-2024", "title": "LLMs4OM: Matching Ontologies with Large Language Models", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2024, "url": "https://arxiv.org/abs/2404.10317", "triage_tier": "core", "triage_score": 110, "sample_claims": [ "LLMs can support ontology matching when prompts include structural context around concepts.", "Retrieval plus matching is a useful architecture for aligning heterogeneous ontologies.", "Ontology matching remains a key semantic interoperability task for multi-source AI systems.", "Retrieval-before-matching helps control LLM cost and candidate scope in ontology matching." ], "sample_chunks": [ "chunk-phase2-llms4om-2024-001", "chunk-phase2-llms4om-2024-002", "chunk-phase2-llms4om-2024-003", "chunk-phase2-llms4om-2024-004" ] }, { "source_id": "phase2-agent-om-2023", "title": "Agent-OM: Leveraging LLM Agents for Ontology Matching", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2023, "url": "https://arxiv.org/abs/2312.00326", "triage_tier": "core", "triage_score": 92, "sample_claims": [ "LLM agents can decompose ontology matching into retrieval, comparison, memory, and tool-use steps.", "Agentic ontology matching is especially promising for complex or few-shot matching tasks.", "Tool boundaries and memory design become part of semantic interoperability systems.", "LLM agents can structure ontology matching as tool-using retrieval and match decisions." ], "sample_chunks": [ "chunk-phase2-agent-om-2023-001", "chunk-phase2-agent-om-2023-002", "chunk-phase2-agent-om-2023-003", "chunk-phase2-agent-om-2023-004" ] }, { "source_id": "p2-llm-ont-007", "title": "Large language models as oracles for instantiating ontologies with domain-specific knowledge", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2025, "url": "https://arxiv.org/abs/2404.04108", "triage_tier": "candidate", "triage_score": 54, "sample_claims": [ "LLMs can populate ontology schemas faster when prompts are structured around classes, properties, and templates.", "Schema guidance reduces but does not eliminate erroneous entities and relations.", "Generated instances are best treated as candidates for expert keep, adjust, discard, or complement decisions." ], "sample_chunks": [ "chunk-p2-llm-ont-007-001", "chunk-p2-llm-ont-007-002", "chunk-p2-llm-ont-007-003", "chunk-p2-llm-ont-007-004" ] }, { "source_id": "p2-llm-ont-030", "title": "Integrating Large Language Models and Knowledge Graphs for Extraction and Validation of Textual Test Data", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2024, "url": "https://arxiv.org/abs/2408.01700", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "LLM extraction becomes more useful when validated against a KG and domain ontology.", "Virtual knowledge graphs can integrate structured and unstructured test data without full physical consolidation.", "Industrial LLM use cases need cost-benefit analysis alongside accuracy." ], "sample_chunks": [ "chunk-p2-llm-ont-030-001", "chunk-p2-llm-ont-030-002", "chunk-p2-llm-ont-030-003", "chunk-p2-llm-ont-030-004" ] }, { "source_id": "phase6-lippolis-2025-llm-assisting-ontology-evaluation", "title": "Large Language Models Assisting Ontology Evaluation", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2025, "url": "https://doi.org/10.1007/978-3-032-09527-5_27", "triage_tier": "candidate", "triage_score": 54, "sample_claims": [ "LLMs can assist competency-question verification at roughly human-user performance levels in the reported setting.", "Hybrid expert-plus-LLM evaluation can reduce ontology evaluation effort while retaining expert control.", "The paper contributes a dataset of competency questions paired with ontologies and ontology stories." ], "sample_chunks": [ "chunk-phase6-lippolis-2025-llm-assisting-ontology-evaluation-001", "chunk-phase6-lippolis-2025-llm-assisting-ontology-evaluation-002", "chunk-phase6-lippolis-2025-llm-assisting-ontology-evaluation-003", "chunk-phase6-lippolis-2025-llm-assisting-ontology-evaluation-004" ] }, { "source_id": "phase6-zhao-2025-llm-ontology-requirements-engineering", "title": "Leveraging Large Language Models for Ontology Requirements Engineering", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2025, "url": "https://doi.org/10.1007/978-3-031-99554-5_40", "triage_tier": "candidate", "triage_score": 54, "sample_claims": [ "LLMs may reduce ambiguity and bias in collaborative ontology requirements elicitation.", "Ontology requirements are a promising intervention point before downstream ontology development.", "Requirements engineering connects user stories and competency questions to ontology scope." ], "sample_chunks": [ "chunk-phase6-zhao-2025-llm-ontology-requirements-engineering-001", "chunk-phase6-zhao-2025-llm-ontology-requirements-engineering-002", "chunk-phase6-zhao-2025-llm-ontology-requirements-engineering-003", "chunk-phase6-zhao-2025-llm-ontology-requirements-engineering-004" ] }, { "source_id": "phase6-lippolis-2025-domain-specific-ontology-generation-llms", "title": "Assessing the Capability of Large Language Models for Domain-Specific Ontology Generation", "bucket": "academic", "quality_signal": "preprint", "year": 2025, "url": "https://arxiv.org/abs/2504.17402", "triage_tier": "candidate", "triage_score": 42, "sample_claims": [], "sample_chunks": [] }, { "source_id": "phase6-kampars-2025-llm-collaborative-ontology-design", "title": "LLM-Supported Collaborative Ontology Design for Data and Knowledge Management Platforms", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2025, "url": "https://doi.org/10.3389/fdata.2025.1676477", "triage_tier": "low_priority", "triage_score": 36, "sample_claims": [], "sample_chunks": [] }, { "source_id": "phase11-oyewale-soru-2026-ontoekg-enterprise-ontology-construction", "title": "LLM-Driven Ontology Construction for Enterprise Knowledge Graphs", "bucket": "academic", "quality_signal": "preprint", "year": 2026, "url": "https://arxiv.org/abs/2602.01276", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [ "Enterprise knowledge graphs require ontologies for semantic governance over heterogeneous data.", "OntoEKG decomposes LLM-driven ontology construction into extraction and entailment modules before serializing standard RDF.", "Reported results show promise while exposing limitations in scope definition and hierarchical reasoning.", "The paper highlights a shortage of comprehensive benchmarks for end-to-end enterprise ontology construction." ], "sample_chunks": [ "chunk-phase11-oyewale-soru-2026-ontoekg-enterprise-ontology-construction-001", "chunk-phase11-oyewale-soru-2026-ontoekg-enterprise-ontology-construction-002", "chunk-phase11-oyewale-soru-2026-ontoekg-enterprise-ontology-construction-003", "chunk-phase11-oyewale-soru-2026-ontoekg-enterprise-ontology-construction-004" ] }, { "source_id": "phase11-soares-wassermann-2026-specific-domain-ontology-construction-llms", "title": "Specific Domain Ontology Construction Using Large Language Models", "bucket": "academic", "quality_signal": "preprint", "year": 2026, "url": "https://arxiv.org/abs/2606.20691", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [ "LLMs can generate generally coherent domain conceptual hierarchies for a specific domain.", "Human expert evaluation found that none of the generated ontologies was completely satisfactory without refinement.", "LLM-based ontology construction should be treated as assisted drafting rather than finished ontology engineering.", "The study reinforces the need for human-in-the-loop evaluation and domain-specific quality criteria." ], "sample_chunks": [ "chunk-phase11-soares-wassermann-2026-specific-domain-ontology-construction-llms-001", "chunk-phase11-soares-wassermann-2026-specific-domain-ontology-construction-llms-002", "chunk-phase11-soares-wassermann-2026-specific-domain-ontology-construction-llms-003", "chunk-phase11-soares-wassermann-2026-specific-domain-ontology-construction-llms-004" ] }, { "source_id": "p2-llm-ont-006", "title": "Large Language Models for Scholarly Ontology Generation: An Extensive Analysis in the Engineering Field", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2025, "url": "https://doi.org/10.1016/j.ipm.2025.104262", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [ "LLMs can classify broader, narrower, same-as, and other topic relations with strong performance under some prompting strategies.", "Smaller or quantized models may be competitive when prompts are tuned for the task.", "Scholarly ontologies need relation-specific evaluation rather than generic text similarity scores.", "LLMs can support relationship classification among scholarly concepts." ], "sample_chunks": [ "chunk-p2-llm-ont-006-001", "chunk-p2-llm-ont-006-002", "chunk-p2-llm-ont-006-003", "chunk-p2-llm-ont-006-004" ] } ] }, { "id": "palantir_operational_ontology", "title": "Palantir: ontology as executable operational layer", "thesis": "Palantir is the clearest commercial case where ontology is presented as an operational substrate: objects and relationships describe the enterprise, while actions, functions, permissions, evals, observability, and MCP tool exposure make it executable by humans and agents.", "source_ids": [ "core-palantir-ontology-overview-2026", "core-palantir-aip-overview-2026", "phase2-pal-ontology-mcp-overview-2026", "phase2-pal-aip-evals-ontology-edits-2026", "phase2-pal-aip-observability-overview-2026", "pal-doc-aip-ethics-governance-2026", "phase3-pal-mcp-installation-2026", "phase3-pal-mcp-security-2026", "phase5-palantir-connecting-agents-decisions-2026", "phase5-palantir-mcp-hub-announcement-2026", "phase5-palantir-foundry-platform-summary-llm-2026" ], "claims": [ "Palantir documents its Ontology as an operational layer over data and models with semantic and kinetic elements.", "AIP is documented as building AI workflows, agents, functions, evals, and applications on top of the Ontology.", "Ontology MCP exposes selected object types, action types, and query functions as MCP tools for external agents.", "AIP Evals, ontology simulations, observability, logs, traces, lineage, permissions, and MCP security docs show a concrete governance architecture for action-capable AI.", "Palantir's current decision-centric narrative extends ontology from representation to decisions, tools, scenarios, and writeback." ], "caveats": [ "Most architecture evidence is official/vendor-authored; label it as Palantir documentation, not independent validation.", "Controls demonstrate design, not necessarily real-world safety, fairness, or public legitimacy.", "Commercial ROI and customer-impact claims must stay separate from technical architecture evidence." ], "retrieval_concepts": [ "palantir_ontology", "palantir_aip", "governed_action", "mcp", "mcp_hub", "decision_lineage", "ontology_governance" ], "sources": [ { "source_id": "core-palantir-ontology-overview-2026", "title": "Ontology building: Overview", "bucket": "palantir", "quality_signal": "official_docs", "year": 2026, "url": "https://www.palantir.com/docs/foundry/ontology/overview", "triage_tier": "core", "triage_score": 130, "sample_claims": [ "Palantir Ontology sits above integrated digital assets and connects them to real-world counterparts.", "The Ontology includes semantic elements and kinetic elements such as actions, functions, and dynamic security.", "The Ontology sits on top of integrated digital assets and connects them to real-world counterparts.", "The Ontology can serve as a digital twin containing semantic elements and kinetic elements." ], "sample_chunks": [ "chunk-core-palantir-ontology-overview-2026-001", "chunk-core-palantir-ontology-overview-2026-002", "chunk-core-palantir-ontology-overview-2026-003", "chunk-core-palantir-ontology-overview-2026-004" ] }, { "source_id": "core-palantir-aip-overview-2026", "title": "AIP overview", "bucket": "palantir", "quality_signal": "official_docs", "year": 2026, "url": "https://palantir.com/docs/foundry/aip/overview", "triage_tier": "core", "triage_score": 124, "sample_claims": [ "AIP builder tools support AI-powered workflows, agents, and functions on top of the Ontology.", "AIP integrates LLMs with platform security, observability, and developer tooling.", "AIP builder tools include AIP Logic, AIP Chatbot Studio, and AIP Evals.", "AIP enables AI workflows, agents, and functions on top of the Ontology and developer toolchain." ], "sample_chunks": [ "chunk-core-palantir-aip-overview-2026-001", "chunk-core-palantir-aip-overview-2026-002", "chunk-core-palantir-aip-overview-2026-003", "chunk-core-palantir-aip-overview-2026-004" ] }, { "source_id": "phase2-pal-ontology-mcp-overview-2026", "title": "Ontology MCP (OMCP) overview", "bucket": "palantir", "quality_signal": "official_docs", "year": 2026, "url": "https://palantir.com/docs/foundry/ontology-mcp/overview", "triage_tier": "core", "triage_score": 118, "sample_claims": [ "Ontology MCP exposes object types, action types, and query functions as MCP tools.", "External agents can discover and use ontology resources to read objects, execute actions, and query data.", "Palantir positions OMCP as a safe read/write bridge between external agents and the Ontology." ], "sample_chunks": [ "chunk-phase2-pal-ontology-mcp-overview-2026-001", "chunk-phase2-pal-ontology-mcp-overview-2026-002", "chunk-phase2-pal-ontology-mcp-overview-2026-003", "chunk-phase2-pal-ontology-mcp-overview-2026-004" ] }, { "source_id": "phase2-pal-aip-evals-ontology-edits-2026", "title": "AIP Evals: Evaluate Ontology edits", "bucket": "palantir", "quality_signal": "official_docs", "year": 2026, "url": "https://palantir.com/docs/foundry/aip-evals/ontology-edits", "triage_tier": "core", "triage_score": 130, "sample_claims": [ "Write-capable AI functions can be evaluated in ontology simulations before affecting real ontology state.", "AIP Evals treats ontology edits as a special case requiring isolation during testing.", "Simulation is a concrete control for non-deterministic LLM workflows with operational side effects.", "AIP Logic functions that result in ontology edits require custom evaluation functions or intermediate parameters." ], "sample_chunks": [ "chunk-phase2-pal-aip-evals-ontology-edits-2026-001", "chunk-phase2-pal-aip-evals-ontology-edits-2026-002", "chunk-phase2-pal-aip-evals-ontology-edits-2026-003", "chunk-phase2-pal-aip-evals-ontology-edits-2026-004" ] }, { "source_id": "phase2-pal-aip-observability-overview-2026", "title": "AIP observability: Overview", "bucket": "palantir", "quality_signal": "official_docs", "year": 2026, "url": "https://palantir.com/docs/foundry/aip-observability/overview", "triage_tier": "core", "triage_score": 142, "sample_claims": [ "AIP observability provides visibility into AIP and Ontology workflow executions through metrics, tracing, logs, and execution history.", "Workflow Lineage integrates observability across applications, workflows, and products built with AIP and the Ontology.", "Runtime observability is part of governing AI workflows, not just offline model evaluation.", "AIP observability provides visibility into AIP and Ontology workflow executions." ], "sample_chunks": [ "chunk-phase2-pal-aip-observability-overview-2026-001", "chunk-phase2-pal-aip-observability-overview-2026-002", "chunk-phase2-pal-aip-observability-overview-2026-003", "chunk-phase2-pal-aip-observability-overview-2026-004" ] }, { "source_id": "pal-doc-aip-ethics-governance-2026", "title": "AI ethics and governance", "bucket": "palantir", "quality_signal": "official_docs", "year": 2026, "url": "https://palantir.com/docs/foundry/aip/ethics-governance", "triage_tier": "core", "triage_score": 86, "sample_claims": [ "AIP Logic tools can delegate tasks to interpretable components rather than relying only on LLM processing.", "Debug views can provide visibility into tool orchestration and handoffs.", "AIP Evals and modeling objectives support testing and governance.", "AIP Evals can evaluate AIP Logic functions across diverse test cases." ], "sample_chunks": [ "chunk-pal-doc-aip-ethics-governance-2026-001", "chunk-pal-doc-aip-ethics-governance-2026-002", "chunk-pal-doc-aip-ethics-governance-2026-003", "chunk-pal-doc-aip-ethics-governance-2026-004" ] }, { "source_id": "phase3-pal-mcp-installation-2026", "title": "Palantir MCP: Installation", "bucket": "palantir", "quality_signal": "official_docs", "year": 2026, "url": "https://palantir.com/docs/foundry/palantir-mcp/installation", "triage_tier": "core", "triage_score": 80, "sample_claims": [ "Palantir MCP uses configured user token permissions.", "External AI systems can read data and metadata from Foundry through AI-friendly APIs.", "Once data is accessed by an external system, governance of its use shifts toward that external system." ], "sample_chunks": [ "chunk-phase3-pal-mcp-installation-2026-001", "chunk-phase3-pal-mcp-installation-2026-002", "chunk-phase3-pal-mcp-installation-2026-003", "chunk-phase3-pal-mcp-installation-2026-004" ] }, { "source_id": "phase3-pal-mcp-security-2026", "title": "Palantir MCP: Security - Data governance", "bucket": "palantir", "quality_signal": "official_docs", "year": 2026, "url": "https://palantir.com/docs/foundry/palantir-mcp/security", "triage_tier": "core", "triage_score": 86, "sample_claims": [ "Palantir MCP provides integration between AI systems and Foundry resources.", "Security and data-governance policies depend on how and where MCP is used.", "MCP use cases require attention to data-flow and security models." ], "sample_chunks": [ "chunk-phase3-pal-mcp-security-2026-001", "chunk-phase3-pal-mcp-security-2026-002", "chunk-phase3-pal-mcp-security-2026-003", "chunk-phase3-pal-mcp-security-2026-004" ] }, { "source_id": "phase5-palantir-connecting-agents-decisions-2026", "title": "Connecting Agents to Decisions", "bucket": "palantir", "quality_signal": "primary_source", "year": 2026, "url": "https://blog.palantir.com/connecting-agents-to-decisions-277dee8ddb40", "triage_tier": "core", "triage_score": 78, "sample_claims": [ "Palantir frames the Ontology as a decision-centric architecture rather than a data-centric schema.", "The blog describes operational decisions as composed of data, logic, action, and security.", "The Ontology is presented as a way to surface enterprise logic assets as AI-ready tools for human-agent workflows.", "The blog claims ontology actions can be staged as scenarios, governed by granular controls, and written back to operational systems." ], "sample_chunks": [ "chunk-phase5-palantir-connecting-agents-decisions-2026-001", "chunk-phase5-palantir-connecting-agents-decisions-2026-002", "chunk-phase5-palantir-connecting-agents-decisions-2026-003", "chunk-phase5-palantir-connecting-agents-decisions-2026-004" ] }, { "source_id": "phase5-palantir-mcp-hub-announcement-2026", "title": "Discover and manage Ontology MCP servers in MCP Hub", "bucket": "palantir", "quality_signal": "official_docs", "year": 2026, "url": "https://www.palantir.com/docs/foundry/announcements/2026-05", "triage_tier": "core", "triage_score": 80, "sample_claims": [ "Ontology MCP servers are discoverable through MCP Hub as of the 2026-05-21 announcement.", "Ontology MCP turns a Developer Console application into an MCP server for external AI agents.", "External AI agents can read object types, execute predefined action types, and run query functions, scoped to configured permissions.", "MCP Hub exposes server configuration details including tools and ontology resources." ], "sample_chunks": [ "chunk-phase5-palantir-mcp-hub-announcement-2026-001", "chunk-phase5-palantir-mcp-hub-announcement-2026-002", "chunk-phase5-palantir-mcp-hub-announcement-2026-003", "chunk-phase5-palantir-mcp-hub-announcement-2026-004" ] }, { "source_id": "phase5-palantir-foundry-platform-summary-llm-2026", "title": "Foundry platform summary for LLMs", "bucket": "palantir", "quality_signal": "official_docs", "year": 2026, "url": "https://www.palantir.com/docs/foundry/getting-started/foundry-platform-summary-llm", "triage_tier": "core", "triage_score": 86, "sample_claims": [ "Ontology MCP exposes application ontology resources as MCP tools for external AI agents.", "External agents can read objects, execute actions, and query data through configured ontology resources.", "Ontology MCP access is restricted through application restrictions and permissions and authenticates through OAuth 2.0 flows.", "The summary lists integrations with desktop agents such as Claude.ai, Microsoft Copilot Studio, Gemini Enterprise, and headless agent frameworks." ], "sample_chunks": [ "chunk-phase5-palantir-foundry-platform-summary-llm-2026-001", "chunk-phase5-palantir-foundry-platform-summary-llm-2026-002", "chunk-phase5-palantir-foundry-platform-summary-llm-2026-003", "chunk-phase5-palantir-foundry-platform-summary-llm-2026-004" ] } ] }, { "id": "enterprise_comparators", "title": "Enterprise comparators: semantic layers, data products, digital twins, and agent memory", "thesis": "Non-Palantir enterprise systems are converging on similar primitives: semantic layers, certified metrics, data products, catalogs, temporal knowledge graphs, digital twins, and governed agent memory.", "source_ids": [ "phase3-llm-ontmem-024", "p2-llm-ont-029", "phase4-enterprise-databricks-genie-semantic-model-2026", "phase4-enterprise-google-looker-semantic-model-gemini-2026", "phase4-enterprise-microsoft-fabric-semantic-model-copilot-2026", "phase4-enterprise-dataworld-ai-context-engine-2026", "phase4-enterprise-dbt-semantic-layer-2026", "phase14-xiao-2018-ontology-based-data-access-survey", "phase14-calvanese-2016-ontop-answering-sparql-relational-databases", "phase15-arenas-2024-morph-kgc-scalable-kg-materialization", "phase14-croissant-ml-ready-dataset-metadata-2024", "phase14-wikidata-free-collaborative-knowledgebase-2014", "phase15-ids-information-model-ontology-2020", "phase15-idsa-semantic-interoperability-data-spaces-2024", "phase15-dataspace-protocol-2025-1", "phase15-gaiax-ontology-compliance-policy-reasoning-2023", "phase16-opc-ua-address-space-model-10000-3", "phase16-etsi-saref-smart-applications-reference-ontology", "phase16-eclass-rdf-owl-product-classification", "phase16-iec-common-data-dictionary-cdd", "phase16-solid-protocol-linked-data-pods-2025", "p2-llm-ont-036", "phase6-odcs-310-open-data-contract-standard", "phase6-odps-41-open-data-product-specification", "phase6-openlineage-docs-specification", "phase6-idta-aas-metamodel-01001", "phase11-qian-2026-brick-dicl-schema-classification", "phase11-shuai-2026-usd-scenes-ontology-grounding-llms", "phase6-xu-2025-a-mem-agentic-memory", "phase3-llm-ontmem-018" ], "claims": [ "Semantic layers can improve natural-language analytics by making metrics, joins, dimensions, and business terms explicit.", "Modern data platforms increasingly package certified data, semantic models, instructions, and curated context for LLM analytics.", "Temporal knowledge graphs and digital twins show why AI memory needs time, events, provenance, and lifecycle semantics.", "Open data-product, data-contract, lineage, and industrial digital-twin specifications show converging machine-readable contracts around operational semantics.", "OBDA and Ontop show a mature academic path for virtual knowledge graphs: users query existing databases through ontology-mediated conceptual layers rather than copying every source into one graph.", "Morph-KGC shows that production-quality knowledge graph construction also needs scalable mapping execution over heterogeneous sources.", "Data-space specifications such as IDS, Dataspace Protocol, and Gaia-X show a federated route where semantic interoperability, usage policies, agreements, credentials, and compliance checks are part of the architecture.", "Industrial semantic standards such as OPC UA, SAREF, eCl@ss, and IEC CDD show that operational AI can inherit mature information models for assets, IoT devices, product classes, and component properties.", "Solid provides a contrasting decentralized linked-data model for permission-aware context and agent memory.", "Croissant and Wikidata show two complementary open comparators: dataset-level ML metadata and public collaborative KG infrastructure.", "Recent Brick-schema and USD-scene work shows the same pattern in buildings, robotics, and simulation: LLMs help map messy operational objects into formal ontology classes, but performance depends on meaningful semantic context and review.", "Agent-memory work is converging on linked, evolving, graph-like memory rather than passive transcript storage.", "Commercial materials are useful adoption evidence but should be separated from peer-reviewed proof." ], "caveats": [ "Vendor documentation often lacks controlled measurement.", "Semantic layers can become stale unless ownership, versioning, validation, and change workflow are explicit." ], "retrieval_concepts": [ "semantic_layer", "ontology_based_data_access", "virtual_knowledge_graph", "knowledge_graph_construction_pipeline", "data_space", "federated_data_sharing", "industrial_semantic_standards", "compliance_policy_reasoning", "ml_dataset_metadata", "open_knowledge_graph", "data_product", "agent_memory", "temporal_knowledge_graph", "digital_twin", "brick_schema", "ontology_grounding" ], "sources": [ { "source_id": "phase3-llm-ontmem-024", "title": "Semantic Layers for Reliable LLM-Powered Data Analytics", "bucket": "academic", "quality_signal": "preprint", "year": 2026, "url": "https://arxiv.org/abs/2604.25149", "triage_tier": "low_priority", "triage_score": 30, "sample_claims": [ "A semantic-layer document can improve LLM analytics accuracy more than switching between comparable frontier model families.", "Reliable text-to-data analysis depends on shared definitions of metrics, dimensions, joins, and business semantics.", "Semantic layers act as governance and interpretation infrastructure for analytical agents." ], "sample_chunks": [ "chunk-phase3-llm-ontmem-024-001", "chunk-phase3-llm-ontmem-024-002", "chunk-phase3-llm-ontmem-024-003", "chunk-phase3-llm-ontmem-024-004" ] }, { "source_id": "p2-llm-ont-029", "title": "Zep: A Temporal Knowledge Graph Architecture for Agent Memory", "bucket": "technical", "quality_signal": "preprint", "year": 2025, "url": "https://arxiv.org/abs/2501.13956", "triage_tier": "low_priority", "triage_score": 30, "sample_claims": [], "sample_chunks": [] }, { "source_id": "phase4-enterprise-databricks-genie-semantic-model-2026", "title": "AI/BI Genie spaces and semantic model documentation", "bucket": "commercial", "quality_signal": "marketing", "year": 2026, "url": "https://docs.databricks.com/en/genie/index.html", "triage_tier": "low_priority", "triage_score": 10, "sample_claims": [], "sample_chunks": [] }, { "source_id": "phase4-enterprise-google-looker-semantic-model-gemini-2026", "title": "Looker semantic model and Gemini in Looker documentation", "bucket": "commercial", "quality_signal": "marketing", "year": 2026, "url": "https://cloud.google.com/looker/docs/semantic-model", "triage_tier": "low_priority", "triage_score": -11, "sample_claims": [], "sample_chunks": [] }, { "source_id": "phase4-enterprise-microsoft-fabric-semantic-model-copilot-2026", "title": "Power BI semantic models and Copilot/Fabric data agent documentation", "bucket": "commercial", "quality_signal": "marketing", "year": 2026, "url": "https://learn.microsoft.com/en-us/fabric/data-science/concept-data-agent", "triage_tier": "low_priority", "triage_score": 16, "sample_claims": [], "sample_chunks": [] }, { "source_id": "phase4-enterprise-dataworld-ai-context-engine-2026", "title": "data.world AI Context Engine and knowledge graph catalog materials", "bucket": "commercial", "quality_signal": "marketing", "year": 2026, "url": "https://data.world/solutions/ai-context-engine", "triage_tier": "low_priority", "triage_score": 10, "sample_claims": [], "sample_chunks": [] }, { "source_id": "phase4-enterprise-dbt-semantic-layer-2026", "title": "dbt Semantic Layer documentation", "bucket": "commercial", "quality_signal": "marketing", "year": 2026, "url": "https://docs.getdbt.com/docs/use-dbt-semantic-layer/dbt-sl", "triage_tier": "low_priority", "triage_score": 16, "sample_claims": [], "sample_chunks": [] }, { "source_id": "phase14-xiao-2018-ontology-based-data-access-survey", "title": "Ontology-Based Data Access: A Survey", "bucket": "academic", "quality_signal": "peer_reviewed_survey", "year": 2018, "url": "https://www.ijcai.org/proceedings/2018/0777.pdf", "triage_tier": "core", "triage_score": 70, "sample_claims": [ "OBDA provides a semantic access layer where users query data repositories through an ontology-mediated conceptual view.", "Mappings connect the ontology layer to underlying relational sources, allowing query rewriting and data access without full consolidation.", "OBDA research connects ontology, databases, query answering, mappings, and heterogeneous data integration." ], "sample_chunks": [ "chunk-phase14-xiao-2018-ontology-based-data-access-survey-001", "chunk-phase14-xiao-2018-ontology-based-data-access-survey-002", "chunk-phase14-xiao-2018-ontology-based-data-access-survey-003", "chunk-phase14-xiao-2018-ontology-based-data-access-survey-004" ] }, { "source_id": "phase14-calvanese-2016-ontop-answering-sparql-relational-databases", "title": "Ontop: Answering SPARQL queries over relational databases", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2016, "url": "https://doi.org/10.3233/SW-160217", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Ontop implements OBDA using an ontology layer, relational sources, and mapping assertions.", "Users can query databases as virtual RDF graphs through SPARQL and OWL/RDF(S)-based conceptual models.", "Virtual knowledge graph systems decouple user-facing semantics from physical storage schemas." ], "sample_chunks": [ "chunk-phase14-calvanese-2016-ontop-answering-sparql-relational-databases-001", "chunk-phase14-calvanese-2016-ontop-answering-sparql-relational-databases-002", "chunk-phase14-calvanese-2016-ontop-answering-sparql-relational-databases-003", "chunk-phase14-calvanese-2016-ontop-answering-sparql-relational-databases-004" ] }, { "source_id": "phase15-arenas-2024-morph-kgc-scalable-kg-materialization", "title": "Morph-KGC: Scalable knowledge graph materialization with mapping partitions", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2024, "url": "https://doi.org/10.3233/SW-223135", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Morph-KGC constructs RDF knowledge graphs from heterogeneous data sources using R2RML and RML.", "Mapping partitions reduce execution time and memory consumption for large knowledge graph materialization tasks.", "Knowledge graph construction needs scalable execution engines in addition to mapping languages." ], "sample_chunks": [ "chunk-phase15-arenas-2024-morph-kgc-scalable-kg-materialization-001", "chunk-phase15-arenas-2024-morph-kgc-scalable-kg-materialization-002", "chunk-phase15-arenas-2024-morph-kgc-scalable-kg-materialization-003", "chunk-phase15-arenas-2024-morph-kgc-scalable-kg-materialization-004" ] }, { "source_id": "phase14-croissant-ml-ready-dataset-metadata-2024", "title": "Croissant: A Metadata Format for ML-Ready Datasets", "bucket": "academic", "quality_signal": "peer_reviewed_survey", "year": 2024, "url": "https://doi.org/10.1145/3650203.3663326", "triage_tier": "core", "triage_score": 70, "sample_claims": [ "Croissant creates a shared metadata representation for ML datasets across tools, frameworks, and platforms.", "The format improves discoverability, portability, interoperability, reproducibility, and responsible-use metadata for datasets.", "Croissant builds on schema.org/JSON-LD and adds ML-specific semantics for resources, record sets, and fields." ], "sample_chunks": [ "chunk-phase14-croissant-ml-ready-dataset-metadata-2024-001", "chunk-phase14-croissant-ml-ready-dataset-metadata-2024-002", "chunk-phase14-croissant-ml-ready-dataset-metadata-2024-003", "chunk-phase14-croissant-ml-ready-dataset-metadata-2024-004" ] }, { "source_id": "phase14-wikidata-free-collaborative-knowledgebase-2014", "title": "Wikidata: a free collaborative knowledgebase", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2014, "url": "https://doi.org/10.1145/2629489", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Wikidata centralizes factual data for Wikipedia and wider reuse as a collaboratively edited knowledgebase.", "Its data model supports multilingual, structured statements that can serve many downstream applications.", "Open collaborative knowledgebases provide reusable entity and relation infrastructure for knowledge graph and AI systems." ], "sample_chunks": [ "chunk-phase14-wikidata-free-collaborative-knowledgebase-2014-001", "chunk-phase14-wikidata-free-collaborative-knowledgebase-2014-002", "chunk-phase14-wikidata-free-collaborative-knowledgebase-2014-003", "chunk-phase14-wikidata-free-collaborative-knowledgebase-2014-004" ] }, { "source_id": "phase15-ids-information-model-ontology-2020", "title": "The International Data Spaces Information Model - An Ontology for Sovereign Exchange of Digital Content", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2020, "url": "https://doi.org/10.1007/978-3-030-62466-8_12", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "The IDS Information Model is an RDFS/OWL ontology for secure and semantically interoperable data spaces.", "It models actors, resources, interactions, contracts, and data exchange concepts needed for sovereign data sharing.", "Data spaces require semantic models in addition to protocols and connectors." ], "sample_chunks": [ "chunk-phase15-ids-information-model-ontology-2020-001", "chunk-phase15-ids-information-model-ontology-2020-002", "chunk-phase15-ids-information-model-ontology-2020-003", "chunk-phase15-ids-information-model-ontology-2020-004" ] }, { "source_id": "phase15-idsa-semantic-interoperability-data-spaces-2024", "title": "Semantic Interoperability in Data Spaces", "bucket": "technical", "quality_signal": "official_docs", "year": 2024, "url": "https://zenodo.org/records/10964377", "triage_tier": "core", "triage_score": 72, "sample_claims": [ "Semantic interoperability in data spaces requires shared information models and understood metadata meanings.", "Data sovereignty and trusted exchange depend on more than transport protocols; participants need shared semantics and governance.", "Data spaces provide a useful lens for cross-organization AI tool and data access governance." ], "sample_chunks": [ "chunk-phase15-idsa-semantic-interoperability-data-spaces-2024-001", "chunk-phase15-idsa-semantic-interoperability-data-spaces-2024-002", "chunk-phase15-idsa-semantic-interoperability-data-spaces-2024-003", "chunk-phase15-idsa-semantic-interoperability-data-spaces-2024-004" ] }, { "source_id": "phase15-dataspace-protocol-2025-1", "title": "Dataspace Protocol 2025-1", "bucket": "technical", "quality_signal": "official_standard", "year": 2025, "url": "https://eclipse-dataspace-protocol-base.github.io/DataspaceProtocol/2025-1", "triage_tier": "core", "triage_score": 76, "sample_claims": [ "The Dataspace Protocol defines interoperable schemas and protocols for federated data sharing.", "It covers publication, agreement negotiation, and data access in dataspace federations.", "Usage control and Web technologies are central to cross-organization data interoperability." ], "sample_chunks": [ "chunk-phase15-dataspace-protocol-2025-1-001", "chunk-phase15-dataspace-protocol-2025-1-002", "chunk-phase15-dataspace-protocol-2025-1-003", "chunk-phase15-dataspace-protocol-2025-1-004" ] }, { "source_id": "phase15-gaiax-ontology-compliance-policy-reasoning-2023", "title": "The Role of Ontologies in Gaia-X", "bucket": "technical", "quality_signal": "official_docs", "year": 2023, "url": "https://gaia-x.eu/the-role-of-ontologies-in-gaia-x", "triage_tier": "core", "triage_score": 72, "sample_claims": [ "Gaia-X uses ontologies for compliance and policy reasoning.", "RDF triples, SHACL shapes, JSON-LD schemas, OWL ontology artifacts, and verifiable credentials support machine-readable trust operations.", "Federated ecosystems need shared semantic descriptions to verify claims and maintain interoperability." ], "sample_chunks": [ "chunk-phase15-gaiax-ontology-compliance-policy-reasoning-2023-001", "chunk-phase15-gaiax-ontology-compliance-policy-reasoning-2023-002", "chunk-phase15-gaiax-ontology-compliance-policy-reasoning-2023-003", "chunk-phase15-gaiax-ontology-compliance-policy-reasoning-2023-004" ] }, { "source_id": "phase16-opc-ua-address-space-model-10000-3", "title": "OPC Unified Architecture Part 3: Address Space Model", "bucket": "technical", "quality_signal": "official_standard", "year": 2025, "url": "https://reference.opcfoundation.org/Core/Part3", "triage_tier": "candidate", "triage_score": 52, "sample_claims": [ "OPC UA uses an address-space information model of nodes, references, attributes, types, objects, variables, methods, and events.", "Industrial interoperability depends on shared machine-readable information models, not only transport protocols.", "Operational AI in manufacturing can use information models as asset and action semantics." ], "sample_chunks": [ "chunk-phase16-opc-ua-address-space-model-10000-3-001", "chunk-phase16-opc-ua-address-space-model-10000-3-002", "chunk-phase16-opc-ua-address-space-model-10000-3-003", "chunk-phase16-opc-ua-address-space-model-10000-3-004" ] }, { "source_id": "phase16-etsi-saref-smart-applications-reference-ontology", "title": "SAREF: Smart Applications REFerence Ontology", "bucket": "technical", "quality_signal": "official_standard", "year": 2025, "url": "https://saref.etsi.org", "triage_tier": "candidate", "triage_score": 64, "sample_claims": [ "SAREF provides a reusable ontology for smart applications and IoT interoperability.", "The ontology models devices, functions, services, measurements, properties, and domain extensions.", "Operational AI over IoT environments needs domain semantics for sensors, actuators, and measurements." ], "sample_chunks": [ "chunk-phase16-etsi-saref-smart-applications-reference-ontology-001", "chunk-phase16-etsi-saref-smart-applications-reference-ontology-002", "chunk-phase16-etsi-saref-smart-applications-reference-ontology-003", "chunk-phase16-etsi-saref-smart-applications-reference-ontology-004" ] }, { "source_id": "phase16-eclass-rdf-owl-product-classification", "title": "eCl@ss in RDF/OWL", "bucket": "technical", "quality_signal": "official_docs", "year": 2025, "url": "https://www.eclass.eu/en/standard/eclassrdfowl", "triage_tier": "core", "triage_score": 72, "sample_claims": [ "Product and service classifications can be represented in RDF/OWL for machine-readable interoperability.", "Standard product semantics support procurement, catalog integration, and supply-chain data exchange.", "Ontology-backed AI in operations benefits from shared product classes and property definitions." ], "sample_chunks": [ "chunk-phase16-eclass-rdf-owl-product-classification-001", "chunk-phase16-eclass-rdf-owl-product-classification-002", "chunk-phase16-eclass-rdf-owl-product-classification-003", "chunk-phase16-eclass-rdf-owl-product-classification-004" ] }, { "source_id": "phase16-iec-common-data-dictionary-cdd", "title": "IEC Common Data Dictionary (IEC CDD)", "bucket": "technical", "quality_signal": "official_standard", "year": 2025, "url": "https://cdd.iec.ch", "triage_tier": "core", "triage_score": 76, "sample_claims": [ "A common data dictionary standardizes product classes, properties, and definitions for electrotechnical domains.", "Industrial interoperability requires shared identifiers and property semantics across tools and organizations.", "Ontology-backed AI agents can use data dictionaries to ground asset and component terms." ], "sample_chunks": [ "chunk-phase16-iec-common-data-dictionary-cdd-001", "chunk-phase16-iec-common-data-dictionary-cdd-002", "chunk-phase16-iec-common-data-dictionary-cdd-003", "chunk-phase16-iec-common-data-dictionary-cdd-004" ] }, { "source_id": "phase16-solid-protocol-linked-data-pods-2025", "title": "Solid Protocol", "bucket": "technical", "quality_signal": "official_docs", "year": 2025, "url": "https://solidproject.org/TR/protocol", "triage_tier": "core", "triage_score": 72, "sample_claims": [ "Solid separates applications from data storage using linked-data resources and access controls.", "Pods provide a decentralized model for personal or organizational data management.", "Permission-aware agent memory and context access can draw on linked-data pod architectures." ], "sample_chunks": [ "chunk-phase16-solid-protocol-linked-data-pods-2025-001", "chunk-phase16-solid-protocol-linked-data-pods-2025-002", "chunk-phase16-solid-protocol-linked-data-pods-2025-003", "chunk-phase16-solid-protocol-linked-data-pods-2025-004" ] }, { "source_id": "p2-llm-ont-036", "title": "Data Products Ontology Specification Version 1.0 beta", "bucket": "technical", "quality_signal": "official_docs", "year": 2025, "url": "https://www.omg.org/spec/DPROD/1.0/Beta1/About-DPROD", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [], "sample_chunks": [] }, { "source_id": "phase6-odcs-310-open-data-contract-standard", "title": "Open Data Contract Standard v3.1.0", "bucket": "technical", "quality_signal": "official_docs", "year": 2026, "url": "https://bitol-io.github.io/open-data-contract-standard/v3.1.0", "triage_tier": "candidate", "triage_score": 66, "sample_claims": [ "Data contracts encode producer-consumer expectations in versionable YAML.", "Contracts combine structural schema with quality and operational commitments.", "The specification is designed for automation and validation." ], "sample_chunks": [ "chunk-phase6-odcs-310-open-data-contract-standard-001", "chunk-phase6-odcs-310-open-data-contract-standard-002", "chunk-phase6-odcs-310-open-data-contract-standard-003", "chunk-phase6-odcs-310-open-data-contract-standard-004" ] }, { "source_id": "phase6-odps-41-open-data-product-specification", "title": "Open Data Product Specification 4.1", "bucket": "technical", "quality_signal": "official_docs", "year": 2026, "url": "https://opendataproducts.org/v4.1", "triage_tier": "candidate", "triage_score": 54, "sample_claims": [ "Data products need a shared, machine-readable structure.", "ODPS includes operational and business metadata beyond dataset schema.", "Version 4.1 adds reference mechanisms for modular governance." ], "sample_chunks": [ "chunk-phase6-odps-41-open-data-product-specification-001", "chunk-phase6-odps-41-open-data-product-specification-002", "chunk-phase6-odps-41-open-data-product-specification-003", "chunk-phase6-odps-41-open-data-product-specification-004" ] }, { "source_id": "phase6-openlineage-docs-specification", "title": "OpenLineage Documentation and Specification", "bucket": "technical", "quality_signal": "official_docs", "year": 2026, "url": "https://openlineage.io/docs", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [ "Lineage should be emitted through shared event metadata rather than custom integrations.", "The core model identifies jobs, runs, and datasets consistently.", "Facets extend the model for system-specific metadata." ], "sample_chunks": [ "chunk-phase6-openlineage-docs-specification-001", "chunk-phase6-openlineage-docs-specification-002", "chunk-phase6-openlineage-docs-specification-003", "chunk-phase6-openlineage-docs-specification-004" ] }, { "source_id": "phase6-idta-aas-metamodel-01001", "title": "Specification of the Asset Administration Shell Part 1: Metamodel, IDTA Number 01001", "bucket": "technical", "quality_signal": "official_docs", "year": 2024, "url": "https://industrialdigitaltwin.org/en?specificationpapers=specification-of-the-asset-administration-shell-part-1-metamodel-idta-number-01001", "triage_tier": "candidate", "triage_score": 54, "sample_claims": [ "AAS represents assets through a formal metamodel and submodels.", "The specification covers identifiers, exchange formats, access control, and mappings to XML, JSON, RDF, AutomationML, and OPC UA.", "The metamodel is intended for partner-to-partner lifecycle information exchange." ], "sample_chunks": [ "chunk-phase6-idta-aas-metamodel-01001-001", "chunk-phase6-idta-aas-metamodel-01001-002", "chunk-phase6-idta-aas-metamodel-01001-003", "chunk-phase6-idta-aas-metamodel-01001-004" ] }, { "source_id": "phase11-qian-2026-brick-dicl-schema-classification", "title": "Brick-DICL: Dynamic In-Context Learning for Automated Brick Schema Classification", "bucket": "academic", "quality_signal": "preprint_benchmark", "year": 2026, "url": "https://arxiv.org/abs/2606.17637", "triage_tier": "candidate", "triage_score": 54, "sample_claims": [ "Brick schema provides a standardized ontology for building systems, but mapping heterogeneous BMS points to classes is difficult.", "Metadata-RAG and class-RAG are used to improve LLM domain knowledge and reduce the large class search space.", "Multi-LLM filtering flags low-confidence classifications for human review.", "The framework targets faster interoperable digital building onboarding while keeping manual verification in the loop." ], "sample_chunks": [ "chunk-phase11-qian-2026-brick-dicl-schema-classification-001", "chunk-phase11-qian-2026-brick-dicl-schema-classification-002", "chunk-phase11-qian-2026-brick-dicl-schema-classification-003", "chunk-phase11-qian-2026-brick-dicl-schema-classification-004" ] }, { "source_id": "phase11-shuai-2026-usd-scenes-ontology-grounding-llms", "title": "From USD Scenes to Knowledge Graphs: Zero-Shot Ontology Grounding with LLMs", "bucket": "academic", "quality_signal": "preprint_benchmark", "year": 2026, "url": "https://arxiv.org/abs/2606.09134", "triage_tier": "candidate", "triage_score": 54, "sample_claims": [ "LLMs can ground scene objects to formal ontology classes in zero-shot settings when descriptive semantic cues are available.", "Scene-graph context such as sibling names and parent paths is a major driver of grounding performance.", "Opaque names and anonymized semantic cues sharply reduce performance, showing limits of LLM grounding without meaningful context.", "Ontology grounding is a bottleneck for constructing KGs from 3D simulation scenes for robot task reasoning." ], "sample_chunks": [ "chunk-phase11-shuai-2026-usd-scenes-ontology-grounding-llms-001", "chunk-phase11-shuai-2026-usd-scenes-ontology-grounding-llms-002", "chunk-phase11-shuai-2026-usd-scenes-ontology-grounding-llms-003", "chunk-phase11-shuai-2026-usd-scenes-ontology-grounding-llms-004" ] }, { "source_id": "phase6-xu-2025-a-mem-agentic-memory", "title": "A-Mem: Agentic Memory for LLM Agents", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2025, "url": "https://openreview.net/forum?id=FiM0M8gcct", "triage_tier": "candidate", "triage_score": 42, "sample_claims": [ "Memory should be actively organized and updated, not only retrieved.", "Zettelkasten-like linked memory improves long-term agent performance in the reported tasks.", "New memories can trigger link generation and evolution of old memories." ], "sample_chunks": [ "chunk-phase6-xu-2025-a-mem-agentic-memory-001", "chunk-phase6-xu-2025-a-mem-agentic-memory-002", "chunk-phase6-xu-2025-a-mem-agentic-memory-003", "chunk-phase6-xu-2025-a-mem-agentic-memory-004" ] }, { "source_id": "phase3-llm-ontmem-018", "title": "Graph-based Agent Memory: Taxonomy, Techniques, and Evaluation", "bucket": "academic", "quality_signal": "peer_reviewed_survey", "year": 2026, "url": "https://arxiv.org/abs/2602.05665", "triage_tier": "candidate", "triage_score": 52, "sample_claims": [ "Graph memory is emerging as a distinct agent-memory family rather than a minor variant of vector retrieval.", "Agent memory evaluation must include update, retrieval, temporal reasoning, access control, and downstream task success.", "Graph memory systems inherit ontology/KG problems such as entity resolution, stale edges, provenance, and schema drift.", "Graphs are a natural substrate for relational, hierarchical, and evolving agent memory." ], "sample_chunks": [ "chunk-phase3-llm-ontmem-018-001", "chunk-phase3-llm-ontmem-018-002", "chunk-phase3-llm-ontmem-018-003", "chunk-phase3-llm-ontmem-018-004" ] } ] }, { "id": "governance_sociotechnical", "title": "Governance and sociotechnical critique", "thesis": "Ontology-backed AI systems are classification infrastructures: they coordinate work, but also encode institutional power, access boundaries, privacy assumptions, and contestable categories.", "source_ids": [ "phase3-star-griesemer-1989-boundary-objects", "phase3-bowker-star-1999-sorting-things-out", "phase3-star-ruhleder-1996-ecology-infrastructure", "phase3-nissenbaum-2010-privacy-context", "phase3-selbst-etal-2019-fairness-abstraction", "phase3-raji-etal-2020-accountability-gap", "phase2-nist-ai-rmf-1-0-2023", "phase2-nist-genai-profile-2024", "phase2-eu-ai-act-2024", "nhs-fdp-contract-explainer-2026", "phase6-nhs-fdp-contracts-finder-2024", "phase6-nhs-ndit-fdp-dpia-2026", "phase6-parliament-rewiring-state-2026", "phase6-ico-dhsc-fdp-foi-2026", "phase6-bmj-morley-zhang-fdp-2023", "phase6-digital-health-ndg-palantir-access-2026", "nhs-fdp-privacy-notice-2024", "bmj-palantir-nhs-contract-2023", "privacy-intl-all-roads-palantir-2021", "amnesty-palantir-human-rights-2020", "oa-https-doi-org-10-1080-01972243-2022-2100851" ], "claims": [ "Ontologies and classifications act as infrastructure and boundary objects, shaping coordination across communities.", "AI governance must address organizational accountability, documentation, logging, oversight, privacy, and context-specific risks.", "NHS FDP is a strong case because official buyer-side documents, privacy notices, reporting, and critiques can be compared.", "Newer public-sector records sharpen the case: procurement notices, DPIAs, ICO decisions, parliamentary scrutiny, and medical governance commentary expose different layers of accountability.", "Public-sector critique focuses less on whether controls exist and more on transparency, legitimacy, procurement scrutiny, trust, rights, and downstream use." ], "caveats": [ "Critique sources should be labeled by type: civil-society report, opinion, journalism, academic article, or official record.", "Do not infer misconduct from the existence of public controversy; use it to frame governance and legitimacy questions." ], "retrieval_concepts": [ "sociotechnical_systems", "boundary_object", "classification_infrastructure", "contextual_integrity", "data_governance", "ai_governance" ], "sources": [ { "source_id": "phase3-star-griesemer-1989-boundary-objects", "title": "Institutional Ecology, 'Translations' and Boundary Objects: Amateurs and Professionals in Berkeley's Museum of Vertebrate Zoology, 1907-39", "bucket": "academic", "quality_signal": "widely_cited", "year": 1989, "url": "https://doi.org/10.1177/030631289019003001", "triage_tier": "candidate", "triage_score": 54, "sample_claims": [ "Boundary objects enable cooperation without requiring full consensus.", "Classification systems, repositories, maps, and forms can coordinate heterogeneous social worlds.", "Standardization and local adaptation coexist in practical scientific infrastructure." ], "sample_chunks": [ "chunk-phase3-star-griesemer-1989-boundary-objects-001", "chunk-phase3-star-griesemer-1989-boundary-objects-002", "chunk-phase3-star-griesemer-1989-boundary-objects-003", "chunk-phase3-star-griesemer-1989-boundary-objects-004" ] }, { "source_id": "phase3-bowker-star-1999-sorting-things-out", "title": "Sorting Things Out: Classification and Its Consequences", "bucket": "books", "quality_signal": "widely_cited", "year": 1999, "url": "https://mitpress.mit.edu/9780262522953/sorting-things-out", "triage_tier": "candidate", "triage_score": 54, "sample_claims": [ "Classifications are embedded in work practices, institutions, and infrastructures.", "Every classification system creates residual categories and consequences for those classified.", "Infrastructure becomes invisible when it works, but failures reveal social and political commitments." ], "sample_chunks": [ "chunk-phase3-bowker-star-1999-sorting-things-out-001", "chunk-phase3-bowker-star-1999-sorting-things-out-002", "chunk-phase3-bowker-star-1999-sorting-things-out-003", "chunk-phase3-bowker-star-1999-sorting-things-out-004" ] }, { "source_id": "phase3-star-ruhleder-1996-ecology-infrastructure", "title": "Steps Toward an Ecology of Infrastructure: Design and Access for Large Information Spaces", "bucket": "academic", "quality_signal": "widely_cited", "year": 1996, "url": "https://doi.org/10.1287/isre.7.1.111", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Infrastructure is not merely built; it becomes infrastructure in use.", "Access, membership, training, and local work practices shape whether shared systems succeed.", "Large information spaces fail when designers ignore local organizational realities." ], "sample_chunks": [ "chunk-phase3-star-ruhleder-1996-ecology-infrastructure-001", "chunk-phase3-star-ruhleder-1996-ecology-infrastructure-002", "chunk-phase3-star-ruhleder-1996-ecology-infrastructure-003", "chunk-phase3-star-ruhleder-1996-ecology-infrastructure-004" ] }, { "source_id": "phase3-nissenbaum-2010-privacy-context", "title": "Privacy in Context: Technology, Policy, and the Integrity of Social Life", "bucket": "books", "quality_signal": "widely_cited", "year": 2010, "url": "https://www.sup.org/books/title?id=8862", "triage_tier": "candidate", "triage_score": 54, "sample_claims": [ "Privacy is about appropriate flows of information, not only secrecy or individual control.", "Social contexts have norms governing actors, attributes, recipients, and transmission principles.", "Technical systems violate privacy when they disrupt contextual information norms." ], "sample_chunks": [ "chunk-phase3-nissenbaum-2010-privacy-context-001", "chunk-phase3-nissenbaum-2010-privacy-context-002", "chunk-phase3-nissenbaum-2010-privacy-context-003", "chunk-phase3-nissenbaum-2010-privacy-context-004" ] }, { "source_id": "phase3-selbst-etal-2019-fairness-abstraction", "title": "Fairness and Abstraction in Sociotechnical Systems", "bucket": "academic", "quality_signal": "widely_cited", "year": 2019, "url": "https://doi.org/10.1145/3287560.3287598", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Fairness failures often arise when technical abstractions ignore social context.", "Systems are embedded in institutions, feedback loops, and power relations.", "Technical metrics cannot substitute for socio-technical analysis of deployment contexts." ], "sample_chunks": [ "chunk-phase3-selbst-etal-2019-fairness-abstraction-001", "chunk-phase3-selbst-etal-2019-fairness-abstraction-002", "chunk-phase3-selbst-etal-2019-fairness-abstraction-003", "chunk-phase3-selbst-etal-2019-fairness-abstraction-004" ] }, { "source_id": "phase3-raji-etal-2020-accountability-gap", "title": "Closing the AI Accountability Gap: Defining an End-to-End Framework for Internal Algorithmic Auditing", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2020, "url": "https://doi.org/10.1145/3351095.3372873", "triage_tier": "candidate", "triage_score": 54, "sample_claims": [ "AI accountability requires process controls, documentation, review, and organizational responsibility.", "Auditing should cover the full lifecycle, not only model performance at one point in time.", "Accountability gaps arise when responsibility is fragmented across teams and stages." ], "sample_chunks": [ "chunk-phase3-raji-etal-2020-accountability-gap-001", "chunk-phase3-raji-etal-2020-accountability-gap-002", "chunk-phase3-raji-etal-2020-accountability-gap-003", "chunk-phase3-raji-etal-2020-accountability-gap-004" ] }, { "source_id": "phase2-nist-ai-rmf-1-0-2023", "title": "Artificial Intelligence Risk Management Framework (AI RMF 1.0)", "bucket": "technical", "quality_signal": "official_standard", "year": 2023, "url": "https://www.nist.gov/itl/ai-risk-management-framework", "triage_tier": "core", "triage_score": 120, "sample_claims": [ "AI risk management requires governance, mapping, measurement, and management across the system lifecycle.", "Trustworthy AI characteristics include validity, reliability, safety, security, resilience, accountability, transparency, explainability, privacy, and fairness.", "Ontology-backed AI systems need explicit risk ownership, measurement, monitoring, and documentation, not only better retrieval." ], "sample_chunks": [ "chunk-phase2-nist-ai-rmf-1-0-2023-001", "chunk-phase2-nist-ai-rmf-1-0-2023-002", "chunk-phase2-nist-ai-rmf-1-0-2023-003", "chunk-phase2-nist-ai-rmf-1-0-2023-004" ] }, { "source_id": "phase2-nist-genai-profile-2024", "title": "Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile", "bucket": "technical", "quality_signal": "official_standard", "year": 2024, "url": "https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.600-1.pdf", "triage_tier": "core", "triage_score": 126, "sample_claims": [ "Generative AI systems create specific risks including confabulation, data privacy leakage, harmful content, and information integrity failures.", "Risk controls should be mapped to specific use contexts rather than treated as generic model properties.", "Ontology-grounded RAG and agent systems need evaluation for grounding, provenance, misuse, privacy, and action consequences." ], "sample_chunks": [ "chunk-phase2-nist-genai-profile-2024-001", "chunk-phase2-nist-genai-profile-2024-002", "chunk-phase2-nist-genai-profile-2024-003", "chunk-phase2-nist-genai-profile-2024-004" ] }, { "source_id": "phase2-eu-ai-act-2024", "title": "Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act)", "bucket": "technical", "quality_signal": "official_standard", "year": 2024, "url": "https://eur-lex.europa.eu/eli/reg/2024/1689/oj", "triage_tier": "core", "triage_score": 120, "sample_claims": [ "High-risk AI systems require risk management, data governance, technical documentation, logging, transparency, human oversight, accuracy, robustness, and cybersecurity controls.", "AI governance is increasingly a legal and operational requirement, not only an ethical preference.", "Operational ontology systems that drive consequential decisions may need auditability, role controls, documentation, and monitoring aligned with legal risk categories." ], "sample_chunks": [ "chunk-phase2-eu-ai-act-2024-001", "chunk-phase2-eu-ai-act-2024-002", "chunk-phase2-eu-ai-act-2024-003", "chunk-phase2-eu-ai-act-2024-004" ] }, { "source_id": "nhs-fdp-contract-explainer-2026", "title": "NHS Federated Data Platform: Contract explainer", "bucket": "technical", "quality_signal": "primary_source", "year": 2026, "url": "https://www.england.nhs.uk/digitaltechnology/nhs-federated-data-platform/security-privacy/contract-explainer", "triage_tier": "candidate", "triage_score": 52, "sample_claims": [ "The contract explainer sets out NHS England's public account of supplier responsibilities and controls.", "It is useful for checking what Palantir is contractually claimed to do and not do.", "The page provides official language for governance, privacy, and public accountability claims.", "A Palantir-led consortium was awarded the NHS FDP contract in November 2023." ], "sample_chunks": [ "chunk-nhs-fdp-contract-explainer-2026-001", "chunk-nhs-fdp-contract-explainer-2026-002", "chunk-nhs-fdp-contract-explainer-2026-003", "chunk-nhs-fdp-contract-explainer-2026-004" ] }, { "source_id": "phase6-nhs-fdp-contracts-finder-2024", "title": "Federated Data Platform and Associated Services", "bucket": "palantir", "quality_signal": "primary_source", "year": 2024, "url": "https://www.contractsfinder.service.gov.uk/Notice/0f8a65b5-23a2-4294-abb1-a7fd8efb3ad0", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "The award notice describes a cloud-based SaaS solution for NHS bodies.", "Palantir Technologies UK Ltd was awarded the FDP contract.", "The notice lists a contract end date of 2027-02-15.", "The notice states a base contract value and forecastable value." ], "sample_chunks": [ "chunk-phase6-nhs-fdp-contracts-finder-2024-001", "chunk-phase6-nhs-fdp-contracts-finder-2024-002", "chunk-phase6-nhs-fdp-contracts-finder-2024-003", "chunk-phase6-nhs-fdp-contracts-finder-2024-004" ] }, { "source_id": "phase6-nhs-ndit-fdp-dpia-2026", "title": "FDP Data Protection Impact Assessment: NDIT Identifiable Version v3.0", "bucket": "palantir", "quality_signal": "primary_source", "year": 2026, "url": "https://www.england.nhs.uk/wp-content/uploads/2025/08/redacted-ndit-nhs-england-fdp-dpia-identifiable-version-v3.0.pdf", "triage_tier": "core", "triage_score": 78, "sample_claims": [ "The DPIA defines Palantir Technologies UK Ltd as Platform Contractor.", "It defines Ontology as mapping datasets and models to object types, properties, link types, and action types.", "It describes national and local FDP instances.", "It flags transparency actions around NDIT." ], "sample_chunks": [ "chunk-phase6-nhs-ndit-fdp-dpia-2026-001", "chunk-phase6-nhs-ndit-fdp-dpia-2026-002", "chunk-phase6-nhs-ndit-fdp-dpia-2026-003", "chunk-phase6-nhs-ndit-fdp-dpia-2026-004" ] }, { "source_id": "phase6-parliament-rewiring-state-2026", "title": "Rewiring the State: Digital Centre of Government", "bucket": "technical", "quality_signal": "primary_source", "year": 2026, "url": "https://committees.parliament.uk/publications/53352/documents/298462/default", "triage_tier": "candidate", "triage_score": 52, "sample_claims": [ "The report argues Palantir should not have such a significant UK public-sector role.", "It calls Palantir's public-sector presence an unacceptable point of weakness.", "It recommends exercising the February 2027 FDP break clause.", "It recommends publishing a costed FDP exit plan by the end of 2026." ], "sample_chunks": [ "chunk-phase6-parliament-rewiring-state-2026-001", "chunk-phase6-parliament-rewiring-state-2026-002", "chunk-phase6-parliament-rewiring-state-2026-003", "chunk-phase6-parliament-rewiring-state-2026-004" ] }, { "source_id": "phase6-ico-dhsc-fdp-foi-2026", "title": "Department of Health & Social Care Decision Notice IC-413447-N9X2", "bucket": "technical", "quality_signal": "primary_source", "year": 2026, "url": "https://ico.org.uk/action-weve-taken/decision-notices/2026/03/ic-413447-n9x2", "triage_tier": "candidate", "triage_score": 40, "sample_claims": [ "The requested information concerned the NHS FDP contract with Palantir.", "The ICO found that the section 35(1)(a) exemption applied.", "No further steps were required." ], "sample_chunks": [ "chunk-phase6-ico-dhsc-fdp-foi-2026-001", "chunk-phase6-ico-dhsc-fdp-foi-2026-002", "chunk-phase6-ico-dhsc-fdp-foi-2026-003", "chunk-phase6-ico-dhsc-fdp-foi-2026-004" ] }, { "source_id": "phase6-bmj-morley-zhang-fdp-2023", "title": "A Controversial New Federated Data Platform for the NHS in England", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2023, "url": "https://pubmed.ncbi.nlm.nih.gov/38030152", "triage_tier": "low_priority", "triage_score": 36, "sample_claims": [ "NHS digital architecture needs improvement.", "FDP controversy turns on trust, scope, value, and lock-in.", "Governance and public legitimacy are central to success." ], "sample_chunks": [ "chunk-phase6-bmj-morley-zhang-fdp-2023-001", "chunk-phase6-bmj-morley-zhang-fdp-2023-002", "chunk-phase6-bmj-morley-zhang-fdp-2023-003", "chunk-phase6-bmj-morley-zhang-fdp-2023-004" ] }, { "source_id": "phase6-digital-health-ndg-palantir-access-2026", "title": "Data Guardian Seeks Clarification on Palantir Patient Data Access", "bucket": "commercial", "quality_signal": "secondary_source", "year": 2026, "url": "https://www.digitalhealth.net/2026/06/data-guardian-seeks-clarification-on-palantir-patient-data-access", "triage_tier": "low_priority", "triage_score": 34, "sample_claims": [ "The National Data Guardian said it was not aware external contractor staff had access to identifiable patient data.", "The NDG identified inconsistency with DPIA materials.", "NHS England said it would work with the NDG and update materials." ], "sample_chunks": [ "chunk-phase6-digital-health-ndg-palantir-access-2026-001", "chunk-phase6-digital-health-ndg-palantir-access-2026-002", "chunk-phase6-digital-health-ndg-palantir-access-2026-003", "chunk-phase6-digital-health-ndg-palantir-access-2026-004" ] }, { "source_id": "nhs-fdp-privacy-notice-2024", "title": "NHS Federated Data Platform privacy notice", "bucket": "technical", "quality_signal": "primary_source", "year": 2024, "url": "https://www.england.nhs.uk/digitaltechnology/nhs-federated-data-platform/security-privacy/nhs-fdp-privacy-notice", "triage_tier": "core", "triage_score": 70, "sample_claims": [ "The FDP processes personal data for specified NHS operational purposes.", "The notice describes who is responsible for processing and who has access.", "The notice presents safeguards, legal bases, and rights information.", "NHS England describes the categories and purposes of data processing in the FDP." ], "sample_chunks": [ "chunk-nhs-fdp-privacy-notice-2024-001", "chunk-nhs-fdp-privacy-notice-2024-002", "chunk-nhs-fdp-privacy-notice-2024-003", "chunk-nhs-fdp-privacy-notice-2024-004" ] }, { "source_id": "bmj-palantir-nhs-contract-2023", "title": "Palantir gets contract to run NHS data platform", "bucket": "technical", "quality_signal": "secondary_source", "year": 2023, "url": "https://www.bmj.com/content/383/bmj.p2752", "triage_tier": "low_priority", "triage_score": 34, "sample_claims": [ "Palantir was reported as winning a major NHS data-platform contract.", "The article documents professional and public concerns about patient data governance.", "Use as secondary reporting and context, not as a technical source for platform architecture." ], "sample_chunks": [ "chunk-bmj-palantir-nhs-contract-2023-001", "chunk-bmj-palantir-nhs-contract-2023-002", "chunk-bmj-palantir-nhs-contract-2023-003", "chunk-bmj-palantir-nhs-contract-2023-004" ] }, { "source_id": "privacy-intl-all-roads-palantir-2021", "title": "All roads lead to Palantir: A review of how the data analytics company operates", "bucket": "technical", "quality_signal": "secondary_source", "year": 2021, "url": "https://privacyinternational.org/sites/default/files/2021-11/All%20roads%20lead%20to%20Palantir%20with%20Palantir%20response%20v3.pdf", "triage_tier": "low_priority", "triage_score": 28, "sample_claims": [ "Palantir's name appears in discussions of predictive policing and public-sector data analytics.", "The report raises concerns about how Palantir platforms can support surveillance or rights-impacting decisions.", "Palantir responses emphasize customer control and safeguards.", "Privacy International argues Palantir's public-sector analytics deployments raise transparency, accountability, and rights concerns." ], "sample_chunks": [ "chunk-privacy-intl-all-roads-palantir-2021-001", "chunk-privacy-intl-all-roads-palantir-2021-002", "chunk-privacy-intl-all-roads-palantir-2021-003", "chunk-privacy-intl-all-roads-palantir-2021-004" ] }, { "source_id": "amnesty-palantir-human-rights-2020", "title": "Failing to Do Rights: The urgent need for Palantir to respect human rights", "bucket": "technical", "quality_signal": "secondary_source", "year": 2020, "url": "https://www.amnestyusa.org/reports/failing-to-do-rights-the-urgent-need-for-palantir-to-respect-human-rights", "triage_tier": "low_priority", "triage_score": 34, "sample_claims": [ "Amnesty argues Palantir deployments can create human-rights risks if due diligence and safeguards are insufficient.", "The report emphasizes accountability, transparency, and downstream use concerns.", "Use as rights-based critique, not as a direct description of Foundry/AIP internals." ], "sample_chunks": [ "chunk-amnesty-palantir-human-rights-2020-001", "chunk-amnesty-palantir-human-rights-2020-002", "chunk-amnesty-palantir-human-rights-2020-003", "chunk-amnesty-palantir-human-rights-2020-004" ] }, { "source_id": "oa-https-doi-org-10-1080-01972243-2022-2100851", "title": "The seer and the seen: Surveying Palantir’s surveillance platform", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2022, "url": "https://www.tandfonline.com/doi/full/10.1080/01972243.2022.2100851", "triage_tier": "candidate", "triage_score": 42, "sample_claims": [ "Palantir is analyzed through the lens of surveillance infrastructure.", "The article discusses government/policing clients and racial justice concerns around predictive policing and analytics.", "It frames Palantir as consequential for democratic and civil-liberties debates.", "The article treats Palantir as a platform that structures what institutions can see, infer, and act on." ], "sample_chunks": [ "chunk-oa-https-doi-org-10-1080-01972243-2022-2100851-001", "chunk-oa-https-doi-org-10-1080-01972243-2022-2100851-002", "chunk-oa-https-doi-org-10-1080-01972243-2022-2100851-003", "chunk-oa-https-doi-org-10-1080-01972243-2022-2100851-004" ] } ] }, { "id": "research_agenda", "title": "Research agenda: from semantic models to accountable AI action", "thesis": "The frontier is not ontology versus LLMs; it is how to build auditable systems where LLMs propose, ontologies constrain, graph memory retrieves, policies authorize, validators check, and humans remain accountable.", "source_ids": [ "phase11-tuan-sanyal-2026-ontology-constrained-neural-reasoning-enterprise-agents", "phase5-liu-2025-ontology-guided-reverse-thinking-kgqa", "phase6-sharma-2025-og-rag", "phase6-xiang-2025-when-to-use-graphs-rag", "phase11-joshi-2026-deontic-policies-agentic-ai-runtime-governance", "phase11-tuan-sanyal-2026-predeployment-assurance-ontology-simulation", "phase11-tuan-sanyal-2026-ontology-constrained-neural-reasoning-enterprise-agents", "phase11-hamed-rocha-2026-biomedical-rag-majority-voting-verification-protocol", "phase3-llm-ontmem-018", "phase2-li-garijo-poveda-2025-llm-oe-review", "phase2-nist-genai-profile-2024", "w3c_shacl_2017", "phase2-w3c-prov-o-2013", "phase16-oasis-xacml-3-policy-access-control-2013", "phase4-std-dpv-2-0-2024", "phase16-w3c-sparql11-federated-query-2013", "phase6-odcs-310-open-data-contract-standard", "phase6-openlineage-docs-specification", "phase4-std-w3c-vc-data-model-2-2025", "phase14-linkml-open-data-modeling-framework-2026", "phase14-sssom-ontological-mappings-2022", "phase14-fair-data-point-metadata-publication-2023", "phase14-hoyt-gyori-2024-o3-guidelines-curated-resources", "phase15-dataspace-protocol-2025-1", "phase15-gaiax-ontology-compliance-policy-reasoning-2023" ], "claims": [ "Ontology-constrained agents are an emerging neurosymbolic direction, but much evidence is still preprint-stage.", "Open problems include ontology drift, graph update validation, provenance of generated triples, permission-aware memory, deletion/forgetting, evaluation, and public accountability.", "Useful future benchmarks should measure end-to-end decision quality, not only answer accuracy or extraction F1.", "The standards and benchmark frontier now points toward executable governance: data contracts, lineage, validation, policy, graph-memory evaluation, and public-sector accountability.", "Machine-actionable schemas, mappings, metadata services, and open infrastructure governance are necessary to make ontology-backed AI maintainable over time.", "Federated data-space protocols and compliance ontologies suggest a research path for cross-organization agent access, agreement negotiation, usage control, and verifiable policy checks.", "Privacy vocabularies, access-control standards, decentralized linked data, and federated query standards suggest a concrete implementation vocabulary for permission-aware retrieval and cross-boundary agent memory.", "New 2026 frontier work frames ontology as runtime policy reasoning, pre-deployment assurance, domain-grounded neurosymbolic validation, and biomedical verification workflow infrastructure." ], "caveats": [ "Keep frontier claims modest and distinguish architecture proposals from mature deployments.", "Nature-style argument should state uncertainty and open questions explicitly." ], "retrieval_concepts": [ "ontology_constrained_reasoning", "runtime_governance", "predeployment_assurance", "semantic_schema", "ontology_mapping_metadata", "mapping_language", "machine_actionable_metadata", "data_space", "compliance_policy_reasoning", "privacy_vocabulary", "access_control_policy", "decentralized_linked_data", "curated_knowledge_infrastructure", "kgqa", "validation", "provenance", "ai_governance", "verifiable_credentials" ], "sources": [ { "source_id": "phase11-tuan-sanyal-2026-ontology-constrained-neural-reasoning-enterprise-agents", "title": "Ontology-Constrained Neural Reasoning in Enterprise Agentic Systems: A Neurosymbolic Architecture for Domain-Grounded AI Agents", "bucket": "academic", "quality_signal": "preprint", "year": 2026, "url": "https://arxiv.org/abs/2604.00555", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [ "Enterprise LLM adoption is constrained by hallucination, domain drift, and weak regulatory enforcement at the reasoning layer.", "A three-layer ontology model can ground roles, domains, and interactions for enterprise agents.", "The paper distinguishes input-side coupling from output-side ontological validation and compliance enforcement.", "Reported experiments suggest ontology-coupled agents can improve metric accuracy and role consistency, with larger value where parametric model knowledge is weaker." ], "sample_chunks": [ "chunk-phase11-tuan-sanyal-2026-ontology-constrained-neural-reasoning-enterprise-agents-001", "chunk-phase11-tuan-sanyal-2026-ontology-constrained-neural-reasoning-enterprise-agents-002", "chunk-phase11-tuan-sanyal-2026-ontology-constrained-neural-reasoning-enterprise-agents-003", "chunk-phase11-tuan-sanyal-2026-ontology-constrained-neural-reasoning-enterprise-agents-004" ] }, { "source_id": "phase5-liu-2025-ontology-guided-reverse-thinking-kgqa", "title": "Ontology-Guided Reverse Thinking Makes Large Language Models Stronger on Knowledge Graph Question Answering", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2025, "url": "https://aclanthology.org/2025.acl-long.741", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Knowledge graph question answering still struggles with multi-hop reasoning when entity vector matching misses abstract question purposes.", "Ontology-Guided Reverse Thinking constructs reasoning paths from purposes back to conditions.", "The method uses LLMs to extract purpose and condition labels, constructs label reasoning paths based on KG ontology, and uses those paths to guide retrieval.", "Experiments on WebQSP and CWQ report state-of-the-art performance and improved LLM KGQA capability." ], "sample_chunks": [ "chunk-phase5-liu-2025-ontology-guided-reverse-thinking-kgqa-001", "chunk-phase5-liu-2025-ontology-guided-reverse-thinking-kgqa-002", "chunk-phase5-liu-2025-ontology-guided-reverse-thinking-kgqa-003", "chunk-phase5-liu-2025-ontology-guided-reverse-thinking-kgqa-004" ] }, { "source_id": "phase6-sharma-2025-og-rag", "title": "OG-RAG: Ontology-Grounded Retrieval-Augmented Generation for Large Language Models", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2025, "url": "https://aclanthology.org/2025.emnlp-main.1674", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Domain ontologies can improve retrieval context construction.", "The method reports gains in accurate fact recall, response correctness, attribution speed, and fact-based reasoning.", "Ontology-grounded RAG targets rule- and workflow-heavy domains where generic RAG underperforms." ], "sample_chunks": [ "chunk-phase6-sharma-2025-og-rag-001", "chunk-phase6-sharma-2025-og-rag-002", "chunk-phase6-sharma-2025-og-rag-003", "chunk-phase6-sharma-2025-og-rag-004" ] }, { "source_id": "phase6-xiang-2025-when-to-use-graphs-rag", "title": "When to Use Graphs in RAG: A Comprehensive Analysis for Graph Retrieval-Augmented Generation", "bucket": "academic", "quality_signal": "preprint_benchmark", "year": 2025, "url": "https://arxiv.org/abs/2506.05690", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [ "GraphRAG does not always outperform vanilla RAG.", "Graph structure helps most under hierarchical retrieval and deeper contextual reasoning needs.", "Evaluation should cover graph construction, retrieval, and generation end to end." ], "sample_chunks": [ "chunk-phase6-xiang-2025-when-to-use-graphs-rag-001", "chunk-phase6-xiang-2025-when-to-use-graphs-rag-002", "chunk-phase6-xiang-2025-when-to-use-graphs-rag-003", "chunk-phase6-xiang-2025-when-to-use-graphs-rag-004" ] }, { "source_id": "phase11-joshi-2026-deontic-policies-agentic-ai-runtime-governance", "title": "Deontic Policies for Runtime Governance of Agentic AI Systems", "bucket": "academic", "quality_signal": "preprint", "year": 2026, "url": "https://arxiv.org/abs/2606.19464", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [ "Tool-using LLM agents create governance needs beyond authentication and access control.", "AgenticRei models permissions, prohibitions, obligations, dispensations, conflict resolution, and ontological reasoning over domain classes.", "The policy engine is evaluated outside the LLM and governs both tool calls and agent-to-agent messages.", "Existing engines such as XACML, Rego, and Cedar are framed as covering only part of the governance structure needed for autonomous agents." ], "sample_chunks": [ "chunk-phase11-joshi-2026-deontic-policies-agentic-ai-runtime-governance-001", "chunk-phase11-joshi-2026-deontic-policies-agentic-ai-runtime-governance-002", "chunk-phase11-joshi-2026-deontic-policies-agentic-ai-runtime-governance-003", "chunk-phase11-joshi-2026-deontic-policies-agentic-ai-runtime-governance-004" ] }, { "source_id": "phase11-tuan-sanyal-2026-predeployment-assurance-ontology-simulation", "title": "Toward Pre-Deployment Assurance for Enterprise AI Agents: Ontology-Grounded Simulation and Trust Certification", "bucket": "academic", "quality_signal": "preprint_benchmark", "year": 2026, "url": "https://arxiv.org/abs/2606.04037", "triage_tier": "candidate", "triage_score": 54, "sample_claims": [ "Pre-deployment verification of enterprise AI agents is a gap between LLM benchmarks and production deployment.", "An Agent Operational Envelope can formalize permissions, domain constraints, safety properties, governance rules, and autonomy levels.", "Ontology-to-scenario generation can derive regulatory, operational, and adversarial tests automatically.", "The framework proposes machine-verifiable Trust Certificates with graduated deployment verdicts." ], "sample_chunks": [ "chunk-phase11-tuan-sanyal-2026-predeployment-assurance-ontology-simulation-001", "chunk-phase11-tuan-sanyal-2026-predeployment-assurance-ontology-simulation-002", "chunk-phase11-tuan-sanyal-2026-predeployment-assurance-ontology-simulation-003", "chunk-phase11-tuan-sanyal-2026-predeployment-assurance-ontology-simulation-004" ] }, { "source_id": "phase11-tuan-sanyal-2026-ontology-constrained-neural-reasoning-enterprise-agents", "title": "Ontology-Constrained Neural Reasoning in Enterprise Agentic Systems: A Neurosymbolic Architecture for Domain-Grounded AI Agents", "bucket": "academic", "quality_signal": "preprint", "year": 2026, "url": "https://arxiv.org/abs/2604.00555", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [ "Enterprise LLM adoption is constrained by hallucination, domain drift, and weak regulatory enforcement at the reasoning layer.", "A three-layer ontology model can ground roles, domains, and interactions for enterprise agents.", "The paper distinguishes input-side coupling from output-side ontological validation and compliance enforcement.", "Reported experiments suggest ontology-coupled agents can improve metric accuracy and role consistency, with larger value where parametric model knowledge is weaker." ], "sample_chunks": [ "chunk-phase11-tuan-sanyal-2026-ontology-constrained-neural-reasoning-enterprise-agents-001", "chunk-phase11-tuan-sanyal-2026-ontology-constrained-neural-reasoning-enterprise-agents-002", "chunk-phase11-tuan-sanyal-2026-ontology-constrained-neural-reasoning-enterprise-agents-003", "chunk-phase11-tuan-sanyal-2026-ontology-constrained-neural-reasoning-enterprise-agents-004" ] }, { "source_id": "phase11-hamed-rocha-2026-biomedical-rag-majority-voting-verification-protocol", "title": "Protocol for evaluating ChatGPT in biomedical association generation and verification using a RAG-enabled, cross-model majority voting workflow", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2026, "url": "https://doi.org/10.1016/j.xpro.2026.104533", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Biomedical generative claims require verification workflows rather than direct acceptance of LLM outputs.", "RAG-enabled cross-model majority voting is proposed as a protocol for evaluating biomedical association generation and verification.", "The work connects to ontology-driven biological graph fact-checking and biomedical knowledge-network verification.", "The protocol is useful as evaluation methodology evidence, not as a general guarantee that voting removes hallucination." ], "sample_chunks": [ "chunk-phase11-hamed-rocha-2026-biomedical-rag-majority-voting-verification-protocol-001", "chunk-phase11-hamed-rocha-2026-biomedical-rag-majority-voting-verification-protocol-002", "chunk-phase11-hamed-rocha-2026-biomedical-rag-majority-voting-verification-protocol-003", "chunk-phase11-hamed-rocha-2026-biomedical-rag-majority-voting-verification-protocol-004" ] }, { "source_id": "phase3-llm-ontmem-018", "title": "Graph-based Agent Memory: Taxonomy, Techniques, and Evaluation", "bucket": "academic", "quality_signal": "peer_reviewed_survey", "year": 2026, "url": "https://arxiv.org/abs/2602.05665", "triage_tier": "candidate", "triage_score": 52, "sample_claims": [ "Graph memory is emerging as a distinct agent-memory family rather than a minor variant of vector retrieval.", "Agent memory evaluation must include update, retrieval, temporal reasoning, access control, and downstream task success.", "Graph memory systems inherit ontology/KG problems such as entity resolution, stale edges, provenance, and schema drift.", "Graphs are a natural substrate for relational, hierarchical, and evolving agent memory." ], "sample_chunks": [ "chunk-phase3-llm-ontmem-018-001", "chunk-phase3-llm-ontmem-018-002", "chunk-phase3-llm-ontmem-018-003", "chunk-phase3-llm-ontmem-018-004" ] }, { "source_id": "phase2-li-garijo-poveda-2025-llm-oe-review", "title": "Large Language Models for Ontology Engineering: A Systematic Literature Review", "bucket": "academic", "quality_signal": "peer_reviewed_survey", "year": 2025, "url": "https://www.semantic-web-journal.net/content/large-language-models-ontology-engineering-systematic-literature-review", "triage_tier": "core", "triage_score": 108, "sample_claims": [ "LLMs are being applied across multiple ontology engineering tasks, but evaluation and reliability remain uneven.", "Ontology engineering with LLMs requires human oversight, task decomposition, and quality controls.", "The literature is shifting from isolated prompting experiments toward workflow-level OE assistance.", "LLMs are being used as ontology engineers, domain experts, and evaluators across the ontology lifecycle." ], "sample_chunks": [ "chunk-phase2-li-garijo-poveda-2025-llm-oe-review-001", "chunk-phase2-li-garijo-poveda-2025-llm-oe-review-002", "chunk-phase2-li-garijo-poveda-2025-llm-oe-review-003", "chunk-phase2-li-garijo-poveda-2025-llm-oe-review-004" ] }, { "source_id": "phase2-nist-genai-profile-2024", "title": "Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile", "bucket": "technical", "quality_signal": "official_standard", "year": 2024, "url": "https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.600-1.pdf", "triage_tier": "core", "triage_score": 126, "sample_claims": [ "Generative AI systems create specific risks including confabulation, data privacy leakage, harmful content, and information integrity failures.", "Risk controls should be mapped to specific use contexts rather than treated as generic model properties.", "Ontology-grounded RAG and agent systems need evaluation for grounding, provenance, misuse, privacy, and action consequences." ], "sample_chunks": [ "chunk-phase2-nist-genai-profile-2024-001", "chunk-phase2-nist-genai-profile-2024-002", "chunk-phase2-nist-genai-profile-2024-003", "chunk-phase2-nist-genai-profile-2024-004" ] }, { "source_id": "w3c_shacl_2017", "title": "Shapes Constraint Language (SHACL)", "bucket": "technical", "quality_signal": "official_standard", "year": 2017, "url": "https://www.w3.org/TR/shacl", "triage_tier": "core", "triage_score": 126, "sample_claims": [ "SHACL separates data graphs from shapes graphs used for validation.", "Constraint validation can check cardinality, datatypes, classes, paths, and custom conditions.", "RDF graph quality can be tested before downstream use.", "SHACL validates RDF graph data against declared constraints." ], "sample_chunks": [ "chunk-w3c_shacl_2017-001", "chunk-w3c_shacl_2017-002", "chunk-w3c_shacl_2017-003", "chunk-w3c_shacl_2017-004" ] }, { "source_id": "phase2-w3c-prov-o-2013", "title": "PROV-O: The PROV Ontology", "bucket": "technical", "quality_signal": "official_standard", "year": 2013, "url": "https://www.w3.org/TR/prov-o", "triage_tier": "core", "triage_score": 120, "sample_claims": [ "PROV-O provides a standard ontology for entities, activities, agents, and their provenance relations.", "Provenance is essential when knowledge graphs are generated, transformed, or used as evidence.", "AI knowledge bases need provenance if generated assertions must remain auditable." ], "sample_chunks": [ "chunk-phase2-w3c-prov-o-2013-001", "chunk-phase2-w3c-prov-o-2013-002", "chunk-phase2-w3c-prov-o-2013-003", "chunk-phase2-w3c-prov-o-2013-004" ] }, { "source_id": "phase16-oasis-xacml-3-policy-access-control-2013", "title": "eXtensible Access Control Markup Language (XACML) Version 3.0", "bucket": "technical", "quality_signal": "official_standard", "year": 2013, "url": "https://docs.oasis-open.org/xacml/3.0/xacml-3.0-core-spec-os-en.html", "triage_tier": "core", "triage_score": 76, "sample_claims": [ "Access-control decisions can be represented as machine-readable policies and request/response evaluations.", "Policies can include obligations and advice that accompany decisions.", "Agentic AI governance needs policy enforcement layers outside the LLM when tools can read or change state." ], "sample_chunks": [ "chunk-phase16-oasis-xacml-3-policy-access-control-2013-001", "chunk-phase16-oasis-xacml-3-policy-access-control-2013-002", "chunk-phase16-oasis-xacml-3-policy-access-control-2013-003", "chunk-phase16-oasis-xacml-3-policy-access-control-2013-004" ] }, { "source_id": "phase4-std-dpv-2-0-2024", "title": "Data Privacy Vocabulary (DPV) Version 2.0", "bucket": "technical", "quality_signal": "official_docs", "year": 2024, "url": "https://www.w3.org/community/reports/dpvcg/CG-FINAL-dpv-20240801", "triage_tier": "core", "triage_score": 72, "sample_claims": [ "Privacy and data processing metadata can be expressed with shared machine-readable vocabularies.", "DPV gives taxonomies for purposes, processing operations, personal data categories, legal bases, rights, risks, and controls.", "AI context and retrieval layers can use privacy vocabularies to constrain use and disclosure." ], "sample_chunks": [ "chunk-phase4-std-dpv-2-0-2024-001", "chunk-phase4-std-dpv-2-0-2024-002", "chunk-phase4-std-dpv-2-0-2024-003", "chunk-phase4-std-dpv-2-0-2024-004" ] }, { "source_id": "phase16-w3c-sparql11-federated-query-2013", "title": "SPARQL 1.1 Federated Query", "bucket": "technical", "quality_signal": "official_standard", "year": 2013, "url": "https://www.w3.org/TR/sparql11-federated-query", "triage_tier": "core", "triage_score": 76, "sample_claims": [ "SPARQL federated query allows portions of a query to be delegated to remote SPARQL services.", "Federated query supports distributed semantic architectures where data remains under separate endpoints.", "Cross-organization ontology-backed AI can use federated query patterns as one retrieval mechanism." ], "sample_chunks": [ "chunk-phase16-w3c-sparql11-federated-query-2013-001", "chunk-phase16-w3c-sparql11-federated-query-2013-002", "chunk-phase16-w3c-sparql11-federated-query-2013-003", "chunk-phase16-w3c-sparql11-federated-query-2013-004" ] }, { "source_id": "phase6-odcs-310-open-data-contract-standard", "title": "Open Data Contract Standard v3.1.0", "bucket": "technical", "quality_signal": "official_docs", "year": 2026, "url": "https://bitol-io.github.io/open-data-contract-standard/v3.1.0", "triage_tier": "candidate", "triage_score": 66, "sample_claims": [ "Data contracts encode producer-consumer expectations in versionable YAML.", "Contracts combine structural schema with quality and operational commitments.", "The specification is designed for automation and validation." ], "sample_chunks": [ "chunk-phase6-odcs-310-open-data-contract-standard-001", "chunk-phase6-odcs-310-open-data-contract-standard-002", "chunk-phase6-odcs-310-open-data-contract-standard-003", "chunk-phase6-odcs-310-open-data-contract-standard-004" ] }, { "source_id": "phase6-openlineage-docs-specification", "title": "OpenLineage Documentation and Specification", "bucket": "technical", "quality_signal": "official_docs", "year": 2026, "url": "https://openlineage.io/docs", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [ "Lineage should be emitted through shared event metadata rather than custom integrations.", "The core model identifies jobs, runs, and datasets consistently.", "Facets extend the model for system-specific metadata." ], "sample_chunks": [ "chunk-phase6-openlineage-docs-specification-001", "chunk-phase6-openlineage-docs-specification-002", "chunk-phase6-openlineage-docs-specification-003", "chunk-phase6-openlineage-docs-specification-004" ] }, { "source_id": "phase4-std-w3c-vc-data-model-2-2025", "title": "Verifiable Credentials Data Model v2.0", "bucket": "technical", "quality_signal": "official_standard", "year": 2025, "url": "https://www.w3.org/TR/vc-data-model-2.0", "triage_tier": "candidate", "triage_score": 46, "sample_claims": [], "sample_chunks": [] }, { "source_id": "phase14-linkml-open-data-modeling-framework-2026", "title": "LinkML: an open data modeling framework", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2026, "url": "https://doi.org/10.1093/gigascience/giaf152", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "LinkML provides an approachable modeling framework for authoring, validating, and sharing structured data schemas.", "The framework connects software-engineering schema needs with linked-data and ontology-ready semantics.", "Typed schemas and validation help make scientific and enterprise data more reusable, interoperable, and less ambiguous for downstream computation." ], "sample_chunks": [ "chunk-phase14-linkml-open-data-modeling-framework-2026-001", "chunk-phase14-linkml-open-data-modeling-framework-2026-002", "chunk-phase14-linkml-open-data-modeling-framework-2026-003", "chunk-phase14-linkml-open-data-modeling-framework-2026-004" ] }, { "source_id": "phase14-sssom-ontological-mappings-2022", "title": "A Simple Standard for Sharing Ontological Mappings (SSSOM)", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2022, "url": "https://doi.org/10.1093/database/baac035", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Ontology mappings require machine-readable metadata that makes mapping semantics, provenance, and uncertainty explicit.", "SSSOM models mappings as pairwise assertions grouped into mapping sets with standard slots.", "FAIR exchange of mappings is a prerequisite for reliable ontology integration, reconciliation, and cross-resource data reuse." ], "sample_chunks": [ "chunk-phase14-sssom-ontological-mappings-2022-001", "chunk-phase14-sssom-ontological-mappings-2022-002", "chunk-phase14-sssom-ontological-mappings-2022-003", "chunk-phase14-sssom-ontological-mappings-2022-004" ] }, { "source_id": "phase14-fair-data-point-metadata-publication-2023", "title": "FAIR Data Point: A FAIR-Oriented Approach for Metadata Publication", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2023, "url": "https://doi.org/10.1162/dint_a_00160", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "FAIR metadata publication requires semantically rich, machine-actionable metadata services.", "FAIR Data Point defines architecture and metadata provision patterns for discovering and searching metadata about digital objects.", "The approach connects linked data, FAIR principles, metadata schemas, and service-level discovery." ], "sample_chunks": [ "chunk-phase14-fair-data-point-metadata-publication-2023-001", "chunk-phase14-fair-data-point-metadata-publication-2023-002", "chunk-phase14-fair-data-point-metadata-publication-2023-003", "chunk-phase14-fair-data-point-metadata-publication-2023-004" ] }, { "source_id": "phase14-hoyt-gyori-2024-o3-guidelines-curated-resources", "title": "The O3 guidelines: open data, open code, and open infrastructure for sustainable curated scientific resources", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2024, "url": "https://doi.org/10.1038/s41597-024-03406-w", "triage_tier": "candidate", "triage_score": 54, "sample_claims": [ "Curated scientific resources can become inaccessible or stale without open data, open code, open infrastructure, and sustainable governance.", "O3 combines technical workflows, social workflows, and progressive governance for community-facing curation.", "Ontology and knowledge-graph infrastructure should be evaluated as maintained resources, not only as one-time datasets." ], "sample_chunks": [ "chunk-phase14-hoyt-gyori-2024-o3-guidelines-curated-resources-001", "chunk-phase14-hoyt-gyori-2024-o3-guidelines-curated-resources-002", "chunk-phase14-hoyt-gyori-2024-o3-guidelines-curated-resources-003", "chunk-phase14-hoyt-gyori-2024-o3-guidelines-curated-resources-004" ] }, { "source_id": "phase15-dataspace-protocol-2025-1", "title": "Dataspace Protocol 2025-1", "bucket": "technical", "quality_signal": "official_standard", "year": 2025, "url": "https://eclipse-dataspace-protocol-base.github.io/DataspaceProtocol/2025-1", "triage_tier": "core", "triage_score": 76, "sample_claims": [ "The Dataspace Protocol defines interoperable schemas and protocols for federated data sharing.", "It covers publication, agreement negotiation, and data access in dataspace federations.", "Usage control and Web technologies are central to cross-organization data interoperability." ], "sample_chunks": [ "chunk-phase15-dataspace-protocol-2025-1-001", "chunk-phase15-dataspace-protocol-2025-1-002", "chunk-phase15-dataspace-protocol-2025-1-003", "chunk-phase15-dataspace-protocol-2025-1-004" ] }, { "source_id": "phase15-gaiax-ontology-compliance-policy-reasoning-2023", "title": "The Role of Ontologies in Gaia-X", "bucket": "technical", "quality_signal": "official_docs", "year": 2023, "url": "https://gaia-x.eu/the-role-of-ontologies-in-gaia-x", "triage_tier": "core", "triage_score": 72, "sample_claims": [ "Gaia-X uses ontologies for compliance and policy reasoning.", "RDF triples, SHACL shapes, JSON-LD schemas, OWL ontology artifacts, and verifiable credentials support machine-readable trust operations.", "Federated ecosystems need shared semantic descriptions to verify claims and maintain interoperability." ], "sample_chunks": [ "chunk-phase15-gaiax-ontology-compliance-policy-reasoning-2023-001", "chunk-phase15-gaiax-ontology-compliance-policy-reasoning-2023-002", "chunk-phase15-gaiax-ontology-compliance-policy-reasoning-2023-003", "chunk-phase15-gaiax-ontology-compliance-policy-reasoning-2023-004" ] } ] }, { "id": "frontier_2026_evidence", "title": "2026 frontier evidence: ontology for hallucination control, graph memory, and agent governance", "thesis": "The newest literature points in a coherent direction: ontology is being used to ground clinical QA, automate enterprise ontology construction, improve graph-based memory retrieval, govern agent tool use, and certify enterprise agents before deployment.", "source_ids": [ "phase11-ali-2026-ontology-grounded-kg-clinical-hallucinations", "phase11-hamed-rocha-2026-biomedical-rag-majority-voting-verification-protocol", "phase11-oyewale-soru-2026-ontoekg-enterprise-ontology-construction", "phase11-soares-wassermann-2026-specific-domain-ontology-construction-llms", "phase11-wu-2026-memgraphrag-memory-multi-agent-graphrag", "phase11-joshi-2026-deontic-policies-agentic-ai-runtime-governance", "phase11-tuan-sanyal-2026-predeployment-assurance-ontology-simulation", "phase11-tuan-sanyal-2026-ontology-constrained-neural-reasoning-enterprise-agents", "phase11-qian-2026-brick-dicl-schema-classification", "phase11-shuai-2026-usd-scenes-ontology-grounding-llms" ], "claims": [ "Peer-reviewed 2026 biomedical sources now directly connect ontology-grounded KGs, RAG-enabled verification, and hallucination mitigation in high-consequence domains.", "Enterprise ontology construction is moving from manual craft toward LLM-assisted extraction, entailment structuring, RDF serialization, and benchmarked evaluation.", "GraphRAG research is shifting from isolated extraction toward memory-aware multi-agent graph construction and hierarchical retrieval.", "Agent governance research is moving toward ontology-backed runtime policy reasoning, operational envelopes, trust certificates, and output-side validation.", "Applied industrial and simulation papers show that ontology grounding/classification can onboard operational objects, but LLM performance depends heavily on semantic cues and human review." ], "caveats": [ "Most sources in this route are preprints; use them to show frontier direction, not settled consensus.", "Do not merge vendor claims and preprint claims into a single proof of deployment effectiveness.", "Biomedical DOI sources are stronger evidence for hallucination/verification themes than enterprise-agent preprints." ], "retrieval_concepts": [ "hallucination_mitigation", "enterprise_ontology_construction", "memgraphrag", "runtime_governance", "predeployment_assurance", "domain_grounded_ai", "ontology_grounding", "brick_schema" ], "sources": [ { "source_id": "phase11-ali-2026-ontology-grounded-kg-clinical-hallucinations", "title": "Ontology-grounded knowledge graphs for mitigating hallucinations in large language models for clinical question answering", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2026, "url": "https://doi.org/10.1016/j.jbi.2026.104993", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Ontology-grounded knowledge graphs are being evaluated as a mitigation strategy for hallucinations in clinical LLM question answering.", "Biomedical QA is a high-consequence setting where semantic grounding, provenance, and domain constraints matter more than open-ended language fluency.", "The article's reference network connects ontology-grounded QA to SPARQL generation, KG-guided RAG, COVID-19 knowledge graphs, and dense passage retrieval." ], "sample_chunks": [ "chunk-phase11-ali-2026-ontology-grounded-kg-clinical-hallucinations-001", "chunk-phase11-ali-2026-ontology-grounded-kg-clinical-hallucinations-002", "chunk-phase11-ali-2026-ontology-grounded-kg-clinical-hallucinations-003", "chunk-phase11-ali-2026-ontology-grounded-kg-clinical-hallucinations-004" ] }, { "source_id": "phase11-hamed-rocha-2026-biomedical-rag-majority-voting-verification-protocol", "title": "Protocol for evaluating ChatGPT in biomedical association generation and verification using a RAG-enabled, cross-model majority voting workflow", "bucket": "academic", "quality_signal": "peer_reviewed", "year": 2026, "url": "https://doi.org/10.1016/j.xpro.2026.104533", "triage_tier": "candidate", "triage_score": 60, "sample_claims": [ "Biomedical generative claims require verification workflows rather than direct acceptance of LLM outputs.", "RAG-enabled cross-model majority voting is proposed as a protocol for evaluating biomedical association generation and verification.", "The work connects to ontology-driven biological graph fact-checking and biomedical knowledge-network verification.", "The protocol is useful as evaluation methodology evidence, not as a general guarantee that voting removes hallucination." ], "sample_chunks": [ "chunk-phase11-hamed-rocha-2026-biomedical-rag-majority-voting-verification-protocol-001", "chunk-phase11-hamed-rocha-2026-biomedical-rag-majority-voting-verification-protocol-002", "chunk-phase11-hamed-rocha-2026-biomedical-rag-majority-voting-verification-protocol-003", "chunk-phase11-hamed-rocha-2026-biomedical-rag-majority-voting-verification-protocol-004" ] }, { "source_id": "phase11-oyewale-soru-2026-ontoekg-enterprise-ontology-construction", "title": "LLM-Driven Ontology Construction for Enterprise Knowledge Graphs", "bucket": "academic", "quality_signal": "preprint", "year": 2026, "url": "https://arxiv.org/abs/2602.01276", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [ "Enterprise knowledge graphs require ontologies for semantic governance over heterogeneous data.", "OntoEKG decomposes LLM-driven ontology construction into extraction and entailment modules before serializing standard RDF.", "Reported results show promise while exposing limitations in scope definition and hierarchical reasoning.", "The paper highlights a shortage of comprehensive benchmarks for end-to-end enterprise ontology construction." ], "sample_chunks": [ "chunk-phase11-oyewale-soru-2026-ontoekg-enterprise-ontology-construction-001", "chunk-phase11-oyewale-soru-2026-ontoekg-enterprise-ontology-construction-002", "chunk-phase11-oyewale-soru-2026-ontoekg-enterprise-ontology-construction-003", "chunk-phase11-oyewale-soru-2026-ontoekg-enterprise-ontology-construction-004" ] }, { "source_id": "phase11-soares-wassermann-2026-specific-domain-ontology-construction-llms", "title": "Specific Domain Ontology Construction Using Large Language Models", "bucket": "academic", "quality_signal": "preprint", "year": 2026, "url": "https://arxiv.org/abs/2606.20691", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [ "LLMs can generate generally coherent domain conceptual hierarchies for a specific domain.", "Human expert evaluation found that none of the generated ontologies was completely satisfactory without refinement.", "LLM-based ontology construction should be treated as assisted drafting rather than finished ontology engineering.", "The study reinforces the need for human-in-the-loop evaluation and domain-specific quality criteria." ], "sample_chunks": [ "chunk-phase11-soares-wassermann-2026-specific-domain-ontology-construction-llms-001", "chunk-phase11-soares-wassermann-2026-specific-domain-ontology-construction-llms-002", "chunk-phase11-soares-wassermann-2026-specific-domain-ontology-construction-llms-003", "chunk-phase11-soares-wassermann-2026-specific-domain-ontology-construction-llms-004" ] }, { "source_id": "phase11-wu-2026-memgraphrag-memory-multi-agent-graphrag", "title": "MemGraphRAG: Memory-based Multi-Agent System for Graph Retrieval-Augmented Generation", "bucket": "academic", "quality_signal": "preprint", "year": 2026, "url": "https://arxiv.org/abs/2606.00610", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [ "GraphRAG can capture structural relationships for complex reasoning, but fragment-level graph extraction can create inconsistent or disconnected graphs.", "Shared memory across agents gives graph construction a global corpus context.", "Memory-aware hierarchical retrieval is proposed to exploit the constructed graph for retrieval-augmented generation.", "The paper reports improvements over baseline models while keeping efficiency comparable." ], "sample_chunks": [ "chunk-phase11-wu-2026-memgraphrag-memory-multi-agent-graphrag-001", "chunk-phase11-wu-2026-memgraphrag-memory-multi-agent-graphrag-002", "chunk-phase11-wu-2026-memgraphrag-memory-multi-agent-graphrag-003", "chunk-phase11-wu-2026-memgraphrag-memory-multi-agent-graphrag-004" ] }, { "source_id": "phase11-joshi-2026-deontic-policies-agentic-ai-runtime-governance", "title": "Deontic Policies for Runtime Governance of Agentic AI Systems", "bucket": "academic", "quality_signal": "preprint", "year": 2026, "url": "https://arxiv.org/abs/2606.19464", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [ "Tool-using LLM agents create governance needs beyond authentication and access control.", "AgenticRei models permissions, prohibitions, obligations, dispensations, conflict resolution, and ontological reasoning over domain classes.", "The policy engine is evaluated outside the LLM and governs both tool calls and agent-to-agent messages.", "Existing engines such as XACML, Rego, and Cedar are framed as covering only part of the governance structure needed for autonomous agents." ], "sample_chunks": [ "chunk-phase11-joshi-2026-deontic-policies-agentic-ai-runtime-governance-001", "chunk-phase11-joshi-2026-deontic-policies-agentic-ai-runtime-governance-002", "chunk-phase11-joshi-2026-deontic-policies-agentic-ai-runtime-governance-003", "chunk-phase11-joshi-2026-deontic-policies-agentic-ai-runtime-governance-004" ] }, { "source_id": "phase11-tuan-sanyal-2026-predeployment-assurance-ontology-simulation", "title": "Toward Pre-Deployment Assurance for Enterprise AI Agents: Ontology-Grounded Simulation and Trust Certification", "bucket": "academic", "quality_signal": "preprint_benchmark", "year": 2026, "url": "https://arxiv.org/abs/2606.04037", "triage_tier": "candidate", "triage_score": 54, "sample_claims": [ "Pre-deployment verification of enterprise AI agents is a gap between LLM benchmarks and production deployment.", "An Agent Operational Envelope can formalize permissions, domain constraints, safety properties, governance rules, and autonomy levels.", "Ontology-to-scenario generation can derive regulatory, operational, and adversarial tests automatically.", "The framework proposes machine-verifiable Trust Certificates with graduated deployment verdicts." ], "sample_chunks": [ "chunk-phase11-tuan-sanyal-2026-predeployment-assurance-ontology-simulation-001", "chunk-phase11-tuan-sanyal-2026-predeployment-assurance-ontology-simulation-002", "chunk-phase11-tuan-sanyal-2026-predeployment-assurance-ontology-simulation-003", "chunk-phase11-tuan-sanyal-2026-predeployment-assurance-ontology-simulation-004" ] }, { "source_id": "phase11-tuan-sanyal-2026-ontology-constrained-neural-reasoning-enterprise-agents", "title": "Ontology-Constrained Neural Reasoning in Enterprise Agentic Systems: A Neurosymbolic Architecture for Domain-Grounded AI Agents", "bucket": "academic", "quality_signal": "preprint", "year": 2026, "url": "https://arxiv.org/abs/2604.00555", "triage_tier": "candidate", "triage_score": 48, "sample_claims": [ "Enterprise LLM adoption is constrained by hallucination, domain drift, and weak regulatory enforcement at the reasoning layer.", "A three-layer ontology model can ground roles, domains, and interactions for enterprise agents.", "The paper distinguishes input-side coupling from output-side ontological validation and compliance enforcement.", "Reported experiments suggest ontology-coupled agents can improve metric accuracy and role consistency, with larger value where parametric model knowledge is weaker." ], "sample_chunks": [ "chunk-phase11-tuan-sanyal-2026-ontology-constrained-neural-reasoning-enterprise-agents-001", "chunk-phase11-tuan-sanyal-2026-ontology-constrained-neural-reasoning-enterprise-agents-002", "chunk-phase11-tuan-sanyal-2026-ontology-constrained-neural-reasoning-enterprise-agents-003", "chunk-phase11-tuan-sanyal-2026-ontology-constrained-neural-reasoning-enterprise-agents-004" ] }, { "source_id": "phase11-qian-2026-brick-dicl-schema-classification", "title": "Brick-DICL: Dynamic In-Context Learning for Automated Brick Schema Classification", "bucket": "academic", "quality_signal": "preprint_benchmark", "year": 2026, "url": "https://arxiv.org/abs/2606.17637", "triage_tier": "candidate", "triage_score": 54, "sample_claims": [ "Brick schema provides a standardized ontology for building systems, but mapping heterogeneous BMS points to classes is difficult.", "Metadata-RAG and class-RAG are used to improve LLM domain knowledge and reduce the large class search space.", "Multi-LLM filtering flags low-confidence classifications for human review.", "The framework targets faster interoperable digital building onboarding while keeping manual verification in the loop." ], "sample_chunks": [ "chunk-phase11-qian-2026-brick-dicl-schema-classification-001", "chunk-phase11-qian-2026-brick-dicl-schema-classification-002", "chunk-phase11-qian-2026-brick-dicl-schema-classification-003", "chunk-phase11-qian-2026-brick-dicl-schema-classification-004" ] }, { "source_id": "phase11-shuai-2026-usd-scenes-ontology-grounding-llms", "title": "From USD Scenes to Knowledge Graphs: Zero-Shot Ontology Grounding with LLMs", "bucket": "academic", "quality_signal": "preprint_benchmark", "year": 2026, "url": "https://arxiv.org/abs/2606.09134", "triage_tier": "candidate", "triage_score": 54, "sample_claims": [ "LLMs can ground scene objects to formal ontology classes in zero-shot settings when descriptive semantic cues are available.", "Scene-graph context such as sibling names and parent paths is a major driver of grounding performance.", "Opaque names and anonymized semantic cues sharply reduce performance, showing limits of LLM grounding without meaningful context.", "Ontology grounding is a bottleneck for constructing KGs from 3D simulation scenes for robot task reasoning." ], "sample_chunks": [ "chunk-phase11-shuai-2026-usd-scenes-ontology-grounding-llms-001", "chunk-phase11-shuai-2026-usd-scenes-ontology-grounding-llms-002", "chunk-phase11-shuai-2026-usd-scenes-ontology-grounding-llms-003", "chunk-phase11-shuai-2026-usd-scenes-ontology-grounding-llms-004" ] } ] } ] }