# Ontology, AI, and Palantir: Core Sources This bibliography is generated from the local triaged knowledge base. Commercial sources are implementation or market evidence, not academic proof. ## 1. AIP observability: Overview - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/aip-observability/overview - Retrieval tags: aip-observability, core, governance, logs, metrics, palantir, tracing, workflow-lineage Official documentation describing AIP observability features for AIP and Ontology workflow executions, including metrics, tracing, logs, execution history, and Workflow Lineage integration. Key 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. - It includes metrics, tracing, logs, and execution history. ## 2. AIP Evals: Evaluate Ontology edits - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/aip-evals/ontology-edits - Retrieval tags: aip-evals, core, custom-evaluation, governed-action, ontology-edits, palantir, simulation, validation, write-capable-agents Official documentation stating that when evaluated Logic functions create, edit, or delete ontology objects, each test case executes in an Ontology simulation so the actual Ontology remains unchanged. Key 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. - Evaluation functions must return Boolean or numeric values, or metrics in a struct. ## 3. Ontology building: Overview - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://www.palantir.com/docs/foundry/ontology/overview - Retrieval tags: actions, core, digital-twin, foundry, kinetic-elements, ontology, operational-layer, palantir, semantic-elements Official documentation defining Palantir Ontology as an operational layer over datasets, virtual tables, and models, with semantic and kinetic elements. Key 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. - Action types and functions define the kinetic parts of organizational change. ## 4. Knowledge Graphs - Authors/Org: Aidan Hogan, Eva Blomqvist, Michael Cochez, Claudia d’Amato, Gerard de Melo, Claudio Gutiérrez, Sabrina Kirrane, José Emilio Labra Gayo - Year: 2021 - Venue/Site: ACM Computing Surveys - Bucket: academic - Quality: peer_reviewed_survey - DOI/Identifier: 10.1145/3447772 - URL: https://dl.acm.org/doi/10.1145/3447772 - Retrieval tags: ai, commercial, construction, core, data-integration, embeddings, knowledge-graph, knowledge-graphs, openalex, rdf In this article, we provide a comprehensive introduction to knowledge graphs, which have recently garnered significant attention from both industry and academia in scenarios that require exploiting diverse, dynamic, large-scale collections of data. After some opening remarks, we motivate and contrast various graph-based data models, as well as languages used to query and validate knowledge graphs. We explain how knowledge can be represented and extracted using a combination of deductive and inductive techniques. We conclude with high-level future research directions for knowledge graphs. Key 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. - Knowledge graph construction and refinement involve extraction, integration, quality control, and curation. ## 5. Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile - Authors/Org: National Institute of Standards and Technology - Year: 2024 - Venue/Site: NIST - Bucket: technical - Quality: official_standard - DOI/Identifier: nist ai 600-1 - URL: https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.600-1.pdf - Retrieval tags: core, evaluation, generative-ai, governance, hallucination, nist, privacy, risk NIST profile applying AI RMF concepts to generative AI, including risks around confabulation, data privacy, harmful content, and information integrity. Key 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. ## 6. RDF 1.1 Concepts and Abstract Syntax - Authors/Org: W3C - Year: 2014 - Venue/Site: W3C - Bucket: technical - Quality: official_standard - DOI/Identifier: w3c recommendation 25 february 2014 - URL: https://www.w3.org/TR/rdf11-concepts - Retrieval tags: core, graph-data, linked-data, rdf, semantic-web, standard, w3c W3C standard defining RDF's graph data model, IRIs, literals, triples, and datasets. Key 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. - RDF datasets support named graphs and graph-based integration. ## 7. Shapes Constraint Language (SHACL) - Authors/Org: W3C RDF Data Shapes Working Group - Year: 2017 - Venue/Site: World Wide Web Consortium Recommendation - Bucket: technical - Quality: official_standard - DOI/Identifier: w3c:rec-shacl-20170720 - URL: https://www.w3.org/TR/shacl - Retrieval tags: constraints, core, data-quality, ontology-governance, rdf, semantic-governance, semantic-web, shacl, validation, w3c Defines SHACL for validating RDF graphs against shapes graphs that express constraints over nodes and properties. Key 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. - Shape validation supports data quality and application-level requirements. ## 8. AIP overview - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/aip/overview - Retrieval tags: agents, ai-platform, aip, aip-evals, aip-logic, automation, core, generative-ai, llm, ontology Official AIP documentation describing AI builder tools for workflows, agents, functions, evals, and applications built on top of the Ontology. Key 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. - AIP integrates generative AI into Palantir's application environment with security, audit, and resource management. ## 9. A Survey on Knowledge Graphs: Representation, Acquisition, and Applications - Authors/Org: Shaoxiong Ji, Shirui Pan, Erik Cambria, Pekka Marttinen, Philip S. Yu - Year: 2021 - Venue/Site: IEEE Transactions on Neural Networks and Learning Systems - Bucket: academic - Quality: peer_reviewed_survey - DOI/Identifier: 10.1109/tnnls.2021.3070843 - URL: https://ieeexplore.ieee.org/document/9416312 - Retrieval tags: applications, core, foundational, kg-completion, knowledge-acquisition, knowledge-graph, ontology_ai, openalex, representation-learning, survey Human knowledge provides a formal understanding of the world. Knowledge graphs that represent structural relations between entities have become an increasingly popular research direction toward cognition and human-level intelligence. In this survey, we provide a comprehensive review of the knowledge graph covering overall research topics about: 1) knowledge graph representation learning; 2) knowledge acquisition and completion; 3) temporal knowledge graph; and 4) knowledge-aware applications and summarize recent breakthroughs and perspective directions to facilitate future research. We propose a full-view categorization and new taxonomies on these topics. Knowledge graph embedding is organized from four aspects of representation space, scoring function, encoding models, and auxiliary information. For knowledge acquisition, especially knowledge graph completion, embedding methods, path inference, and logical rule reasoning are reviewed. We further explore several emerging topics, including metarelational learning, commonsense reasoning, and temporal knowledge graphs. To facilitate future research on knowledge graphs, we also provide a curated collection of data sets and open-source libraries on different tasks. In the end, we have a thorough outlook on several promising research directions. Key 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. - Applications include question answering, recommendation, and information retrieval. ## 10. Artificial Intelligence Risk Management Framework (AI RMF 1.0) - Authors/Org: National Institute of Standards and Technology - Year: 2023 - Venue/Site: NIST - Bucket: technical - Quality: official_standard - DOI/Identifier: nist ai rmf 1.0 - URL: https://www.nist.gov/itl/ai-risk-management-framework - Retrieval tags: ai-risk-management, core, evaluation, governance, nist, policy, trustworthy-ai Official NIST framework for managing AI risks through govern, map, measure, and manage functions across the AI lifecycle. Key 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. ## 11. PROV-O: The PROV Ontology - Authors/Org: W3C Provenance Working Group - Year: 2013 - Venue/Site: W3C - Bucket: technical - Quality: official_standard - URL: https://www.w3.org/TR/prov-o - Retrieval tags: audit, core, lineage, owl, prov-o, provenance, trustworthy-ai, w3c W3C Recommendation expressing the PROV data model as an OWL2 ontology for representing provenance across systems and contexts. Key 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. ## 12. Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act) - Authors/Org: European Parliament and Council of the European Union - Year: 2024 - Venue/Site: Official Journal of the European Union - Bucket: technical - Quality: official_standard - DOI/Identifier: regulation (eu) 2024/1689 - URL: https://eur-lex.europa.eu/eli/reg/2024/1689/oj - Retrieval tags: core, eu-ai-act, governance, human-oversight, logging, public-sector, regulation, risk-management EU regulation establishing rules for AI systems, including risk classification, documentation, transparency, human oversight, logging, post-market monitoring, and governance requirements. Key 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. ## 13. Retrieval-Augmented Generation for Large Language Models: A Survey - Authors/Org: Yunfan Gao; Yun Xiong; Xinyu Gao; Kangxiang Jia; Jinliu Pan; Yuxi Bi; Yi Dai; Jiawei Sun; Meng Wang; Haofen Wang - Year: 2023 - Venue/Site: arXiv - Bucket: academic - Quality: peer_reviewed_survey - DOI/Identifier: arxiv:2312.10997 - URL: https://arxiv.org/abs/2312.10997 - Retrieval tags: advanced-rag, core, evaluation, llm, rag, retrieval, retrieval-pipeline, survey Survey of RAG paradigms, retrieval, augmentation, generation, evaluation, and applications for LLMs. Key claims: - RAG systems depend on retrieval quality, knowledge organization, and generation control. - Advanced RAG includes query transformation, reranking, iterative retrieval, and evaluation. - RAG reduces reliance on static parametric memory and supports external evidence. - RAG quality depends on data indexing, retrieval, reranking, and generation alignment. - Evaluation must include retrieval quality, answer correctness, faithfulness, and robustness. ## 14. Unifying Large Language Models and Knowledge Graphs: A Roadmap - Authors/Org: Shirui Pan, Linhao Luo, Yufei Wang, Chen Chen, Jiapu Wang, Xindong Wu - Year: 2023 - Venue/Site: arXiv - Bucket: academic - Quality: peer_reviewed_survey - DOI/Identifier: 10.1109/tkde.2024.3352100 - URL: https://arxiv.org/abs/2306.08302 - Retrieval tags: arxiv, core, grounding, hybrid-ai, knowledge-graph, large-language-models, llm, llm-kg, ontology_ai, reasoning Large language models (LLMs), such as ChatGPT and GPT4, are making new waves in the field of natural language processing and artificial intelligence, due to their emergent ability and generalizability. However, LLMs are black-box models, which often fall short of capturing and accessing factual knowledge. In contrast, Knowledge Graphs (KGs), Wikipedia and Huapu for example, are structured knowledge models that explicitly store rich factual knowledge. KGs can enhance LLMs by providing external knowledge for inference and interpretability. Meanwhile, KGs are difficult to construct and evolving by nature, which challenges the existing methods in KGs to generate new facts and represent unseen knowledge. Therefore, it is complementary to unify LLMs and KGs together and simultaneously leverage their advantages. In this article, we present a forward-looking roadmap for the unification of LLMs and KGs. Our roadmap consists of three general frameworks, namely, 1) KG-enhanced LLMs, which incorporate KGs during the pre-training and inference phases of LLMs, or for the purpose of enhancing understanding of the knowledge learned by LLMs; 2) LLM-augmented KGs, that leverage LLMs for different KG tasks such as embedding, completion, construction, graph-to-text generation, and question answering; and 3) Synergized LLMs + KGs, in which LLMs and KGs play equal roles and work in a mutually beneficial way to enhance both LLMs and KGs for bidirectional reasoning driven by both data and knowledge. We review and summarize existing efforts within these three frameworks in our roadmap and pinpoint their future research directions. Key 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. - LLMs can help construct and query KGs, while KGs can ground and constrain LLMs. ## 15. Ontology MCP (OMCP) overview - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/ontology-mcp/overview - Retrieval tags: action-type, agents, core, mcp, object-type, ontology-mcp, palantir, tool-use Official documentation describing Ontology MCP as a Developer Console feature that exposes application ontology resources as MCP tools for external AI agents and systems. Key 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. ## 16. From Local to Global: A Graph RAG Approach to Query-Focused Summarization - Authors/Org: Darren Edge; Ha Trinh; Newman Cheng; Joshua Bradley; Alex Chao; Apurva Mody; Steven Truitt; Jonathan Larson - Year: 2024 - Venue/Site: arXiv / Microsoft Research - Bucket: academic - Quality: primary_source - DOI/Identifier: arxiv:2404.16130 - URL: https://arxiv.org/abs/2404.16130 - Retrieval tags: community-summary, core, enterprise-search, evaluation, graphrag, knowledge-graph, microsoft, private-data, provenance, query-focused-summarization Presents GraphRAG, using graph-based communities and summaries to support global question answering over private datasets. Key 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. - Graph construction quality is a key dependency. ## 17. ISO/IEC 42001:2023 Artificial intelligence - Management system - Authors/Org: ISO/IEC - Year: 2023 - Venue/Site: ISO/IEC - Bucket: technical - Quality: official_standard - DOI/Identifier: iso/iec 42001:2023 - URL: https://www.iso.org/standard/81230.html - Retrieval tags: ai-management-system, core, governance, iso, lifecycle, policy, risk-management International standard specifying requirements for establishing, implementing, maintaining, and continually improving an AI management system within organizations. Key claims: - AI management systems require organizational processes for responsibility, risk, lifecycle controls, monitoring, and continual improvement. - Governance must be embedded in management systems, not left to individual model or prompt choices. - Ontology and knowledge graph programs should be governed as AI infrastructure when they support automated decisions or actions. ## 18. SKOS Simple Knowledge Organization System Reference - Authors/Org: W3C Semantic Web Deployment Working Group - Year: 2009 - Venue/Site: World Wide Web Consortium Recommendation - Bucket: technical - Quality: official_standard - DOI/Identifier: w3c rec-skos-reference-20090818 - URL: https://www.w3.org/TR/skos-reference - Retrieval tags: controlled-vocabulary, core, retrieval, semantic-web, skos, standard, taxonomy, w3c Defines SKOS as a model for thesauri, classification schemes, taxonomies, subject headings, and other knowledge organization systems. Key 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. - SKOS is lighter-weight than OWL but valuable for indexing, concept schemes, labels, and semantic retrieval. ## 19. SPARQL 1.1 Query Language - Authors/Org: W3C SPARQL Working Group - Year: 2013 - Venue/Site: W3C - Bucket: technical - Quality: official_standard - DOI/Identifier: w3c:rec-sparql11-query-20130321 - URL: https://www.w3.org/TR/sparql11-query - Retrieval tags: core, graph-query, knowledge-graph, query, query-language, rdf, semantic-web, sparql, standard, standards W3C standard query language for RDF graphs, including graph pattern matching and result forms. Key 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. - Standard graph querying is necessary for interoperable RDF systems. ## 20. Toward principles for the design of ontologies used for knowledge sharing? - Authors/Org: Thomas Gruber - Year: 1995 - Venue/Site: International Journal of Human-Computer Studies - Bucket: academic - Quality: peer_reviewed_seminal - DOI/Identifier: 10.1006/ijhc.1995.1081 - URL: https://doi.org/10.1006/ijhc.1995.1081 - Retrieval tags: clarity, coherence, core, foundational, gruber, knowledge-sharing, minimal-commitment, ontology-design, openalex Develops design principles for shareable ontologies, including clarity, coherence, extendibility, minimal encoding bias, and minimal ontological commitment. Key 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. - Minimal ontological commitment can improve reuse across applications. ## 21. DBpedia – A large-scale, multilingual knowledge base extracted from Wikipedia - Authors/Org: Jens Lehmann, Robert Isele, Max Jakob, Anja Jentzsch, Dimitris Kontokostas, Pablo N. Mendes, Sebastian Hellmann, Mohamed Morsey - Year: 2015 - Venue/Site: Semantic Web - Bucket: academic - Quality: peer_reviewed - DOI/Identifier: 10.3233/sw-140134 - URL: https://doi.org/10.3233/SW-140134 - Retrieval tags: commercial, core, dbpedia, foundational, knowledge-graph, linked-data, openalex The DBpedia community project extracts structured, multilingual knowledge from Wikipedia and makes it freely available on the Web using Semantic Web and Linked Data technologies. The project extracts knowledge from 111 different language editions of Key claims: - Large-scale structured knowledge can be extracted from semi-structured community resources. - Linked Data enables interconnection across datasets. ## 22. From RAG to Memory: Non-Parametric Continual Learning for Large Language Models - Authors/Org: Bernal Jimenez Gutierrez; Yiheng Shu; Weijian Qi; Sizhe Zhou; Yu Su - Year: 2025 - Venue/Site: ICML 2025 / PMLR - Bucket: academic - Quality: peer_reviewed - DOI/Identifier: arxiv:2502.14802 - URL: https://arxiv.org/abs/2502.14802 - Retrieval tags: agent-memory, continual-learning, core, graph-rag, hipporag, knowledge-graph, memory, non-parametric-memory, rag Presents HippoRAG 2, extending graph-based RAG toward non-parametric continual learning with improved factual, sense-making, and associative memory. Key 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. - Continual memory requires updates and retrieval consistency across sessions. ## 23. LLMs4OM: Matching Ontologies with Large Language Models - Authors/Org: Hamed Babaei Giglou; Jennifer D'Souza; Felix Engel; Soren Auer - Year: 2024 - Venue/Site: arXiv / ESWC - Bucket: academic - Quality: peer_reviewed - DOI/Identifier: arxiv:2404.10317 - URL: https://arxiv.org/abs/2404.10317 - Retrieval tags: core, eswc-2024, interoperability, llm, ontology-matching, retrieval, schema-alignment, semantic-integration Evaluates large language models for ontology matching using retrieval and matching modules across concept, parent, and children representations over multiple OM datasets. Key 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. - Parent and child context can improve concept matching beyond label-only prompts. ## 24. Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks - Authors/Org: Patrick Lewis; Ethan Perez; Aleksandra Piktus; Fabio Petroni; Vladimir Karpukhin; Naman Goyal; Heinrich Kuttler; Mike Lewis; Wen-tau Yih; Tim Rocktaschel; Sebastian Riedel; Douwe Kiela - Year: 2020 - Venue/Site: NeurIPS - Bucket: academic - Quality: peer_reviewed_seminal - DOI/Identifier: arxiv:2005.11401 - URL: https://arxiv.org/abs/2005.11401 - Retrieval tags: core, evidence, external-memory, grounding, knowledge-intensive, knowledge-intensive-nlp, llm, rag, retrieval, retrieval-augmented-generation Introduces RAG models that combine parametric generation with non-parametric retrieval for knowledge-intensive NLP. Key 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. - Generated answers depend on retriever quality and evidence selection. ## 25. The Semantic Web - Authors/Org: Tim Berners-Lee; James Hendler; Ora Lassila - Year: 2001 - Venue/Site: Scientific American - Bucket: technical - Quality: widely_cited - DOI/Identifier: 10.1038/scientificamerican0501-34 - URL: https://doi.org/10.1038/scientificamerican0501-34 - Retrieval tags: agents, core, linked-data, machine-readable-data, metadata, semantic-web Landmark article presenting a vision of Web data with machine-interpretable meaning through agents, ontologies, and shared semantics. Key claims: - The Semantic Web aims to make Web content processable by machines through explicit semantics. - Ontologies allow agents and services to interpret data across sources. - The Web can be extended with structured meaning that software agents can process. - Ontologies and metadata help machines interpret data from diverse sources. - Machine-actionable semantics can support automated services and coordination. ## 26. Knowledge graph refinement: A survey of approaches and evaluation methods - Authors/Org: Heiko Paulheim - Year: 2016 - Venue/Site: Semantic Web - Bucket: academic - Quality: peer_reviewed_survey - DOI/Identifier: 10.3233/sw-160218 - URL: https://doi.org/10.3233/SW-160218 - Retrieval tags: core, foundational, knowledge-graph, openalex, quality, refinement, survey In the recent years, different Web knowledge graphs, both free and commercial, have been created. While Google coined the term “Knowledge Graph” in 2012, there are also a few openly available knowledge graphs, with DBpedia, YAGO, and Freebase being a Key claims: - KG refinement covers completion, error detection, type assertion, and relation prediction. - Evaluation must distinguish different refinement tasks and data assumptions. ## 27. Information technology - Top-level ontologies - Part 2: Basic Formal Ontology (BFO) - Authors/Org: ISO/IEC - Year: 2021 - Venue/Site: ISO/IEC - Bucket: technical - Quality: official_standard - DOI/Identifier: iso/iec 21838-2:2021 - URL: https://www.iso.org/standard/74572.html - Retrieval tags: bfo, core, interoperability, iso-21838, standard, top-level-ontology, upper-ontology International standard specifying Basic Formal Ontology as a top-level ontology for interoperable domain ontologies. Key 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. - BFO provides high-level categories for entities, continuants, occurrents, qualities, roles, and relations. ## 28. Large Language Models for Ontology Engineering: A Systematic Literature Review - Authors/Org: J. Li; D. Garijo; M. Poveda-Villalon - Year: 2025 - Venue/Site: Semantic Web Journal preprint - Bucket: academic - Quality: peer_reviewed_survey - URL: https://www.semantic-web-journal.net/content/large-language-models-ontology-engineering-systematic-literature-review - Retrieval tags: benchmarks, core, human-in-the-loop, llm, ontology-engineering, ontology-governance, survey, systematic-review, validation Systematic review of how large language models are being applied across ontology engineering tasks, including generation, refinement, alignment, evaluation, and human-in-the-loop workflows. Key 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. - Current studies lack stable task definitions, shared benchmarks, and reproducible evaluation protocols. ## 29. OWL 2 Web Ontology Language Document Overview - Authors/Org: W3C OWL Working Group - Year: 2012 - Venue/Site: W3C - Bucket: technical - Quality: official_standard - DOI/Identifier: w3c:rec-owl2-overview-20121211 - URL: https://www.w3.org/TR/owl2-overview - Retrieval tags: core, description-logic, knowledge-representation, ontology, ontology-language, owl, reasoning, semantic-web, standard, standards W3C overview of OWL 2, a Semantic Web language for classes, properties, individuals, axioms, and reasoning profiles. Key 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. - OWL enables automated reasoning such as classification and consistency checking. ## 30. The Description Logic Handbook: Theory, Implementation, and Applications - Authors/Org: Franz Baader; Diego Calvanese; Deborah McGuinness; Daniele Nardi; Peter Patel-Schneider - Year: 2003 - Venue/Site: Cambridge University Press - Bucket: books - Quality: scholarly_book - DOI/Identifier: isbn 9780521781763 - URL: https://www.cambridge.org/core/books/description-logic-handbook - Retrieval tags: book, core, description-logic, owl, reasoning Reference handbook on description logics, the formal basis of many ontology languages including OWL. Key claims: - Description logics balance expressivity and decidable reasoning. - Ontology languages can be grounded in formal semantics and automated inference. ## 31. Formal Ontology in Information Systems - Authors/Org: Nicola Guarino - Year: 1998 - Venue/Site: FOIS - Bucket: academic - Quality: peer_reviewed_seminal - DOI/Identifier: fois 1998 - URL: https://dl.acm.org/doi/10.5555/521720.521722 - Retrieval tags: conceptual-modeling, core, formal-ontology, information-systems Positions formal ontology as a foundation for information systems, conceptual modeling, and shared meaning across systems. Key claims: - Formal ontology helps clarify conceptual distinctions in information systems. - Ontology is useful when data structures need meaning-preserving integration. ## 32. G-Retriever: Retrieval-Augmented Generation for Textual Graph Understanding and Question Answering - Authors/Org: Xiaoxin He; Yijun Tian; Yifei Sun; Nitesh V. Chawla; Thomas Laurent; Yann LeCun; Xavier Bresson; Bryan Hooi - Year: 2024 - Venue/Site: NeurIPS 2024 / arXiv - Bucket: academic - Quality: peer_reviewed - DOI/Identifier: arxiv:2402.07630 - URL: https://arxiv.org/abs/2402.07630 - Retrieval tags: core, g-retriever, graph-qa, graph-rag, hallucination, knowledge-graph, rag, retrieval, subgraph-retrieval, textual-graphs Develops a RAG framework for real-world textual graphs and graph question answering, including graph-relevant retrieval via Prize-Collecting Steiner Tree optimization. Key claims: - Graph question answering needs retrieval over graph structure, not only document chunks. - Optimizing subgraph selection can reduce hallucination and fit graph evidence into the LLM context window. - Textual graphs are a practical setting where KG reasoning and LLM generation meet. - Graph QA needs retrieval over both topology and node text. - Subgraph retrieval can reduce context size while preserving structural evidence. ## 33. Handbook on Ontologies - Authors/Org: Steffen Staab; Rudi Studer - Year: 2009 - Venue/Site: Springer - Bucket: books - Quality: widely_cited - DOI/Identifier: 10.1007/978-3-540-92673-3 - URL: https://doi.org/10.1007/978-3-540-92673-3 - Retrieval tags: book, core, handbook, mapping, matching, ontology-engineering, ontology-learning, reasoning, semantic-web Edited handbook covering ontology languages, engineering methods, ontology learning, matching, reasoning, and applications. Key claims: - Ontology engineering combines formal languages, methods, tools, and application practices. - Ontology lifecycle includes modeling, alignment, learning, evaluation, and reuse. - Ontology research spans formal foundations, engineering methods, languages, and applications. - Evaluation, alignment, reasoning, and lifecycle management are central ontology concerns. - Ontology systems require both conceptual rigor and implementation tooling. ## 34. HippoRAG: Neurobiologically Inspired Long-Term Memory for Large Language Models - Authors/Org: Liang Wang; Nan Yang; Furu Wei; et al. - Year: 2024 - Venue/Site: arXiv - Bucket: academic - Quality: peer_reviewed - DOI/Identifier: arxiv:2405.14831 - URL: https://arxiv.org/abs/2405.14831 - Retrieval tags: agent-memory, associative-memory, core, graph-retrieval, hipporag, knowledge-graph, llm, long-term-memory, multi-hop, multi-hop-qa Presents a long-term memory retrieval framework for LLMs inspired by hippocampal indexing, using graph-based associations for multi-hop retrieval. Key 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. - Personalized PageRank over graph memory can reduce cost and improve multi-hop retrieval compared with iterative retrieval. ## 35. Knowledge Representation: Logical, Philosophical, and Computational Foundations - Authors/Org: John F. Sowa - Year: 2000 - Venue/Site: Brooks/Cole - Bucket: books - Quality: widely_cited - DOI/Identifier: isbn 9780534949655 - URL: https://www.jfsowa.com/krbook - Retrieval tags: ai-foundations, book, conceptual-graphs, core, formal-semantics, knowledge-representation, logic, ontology, phase3 Comprehensive treatment of logic, ontology, conceptual graphs, philosophy, and computation for knowledge representation. Key claims: - Knowledge representation sits between logic, philosophy, linguistics, and computation. - Ontological categories influence how systems model and reason about the world. - Knowledge representation requires choices about logic, ontology, language, and computation. - No single representation captures every purpose; expressive power and computational tractability trade off. - Conceptual structures mediate between natural language, databases, and reasoning systems. ## 36. Ontologies: Principles, Methods and Applications - Authors/Org: Mike Uschold; Michael Gruninger - Year: 1996 - Venue/Site: Knowledge Engineering Review - Bucket: academic - Quality: peer_reviewed_seminal - DOI/Identifier: 10.1017/s0269888900007797 - URL: https://doi.org/10.1017/S0269888900007797 - Retrieval tags: competency-questions, core, evaluation, methodology, ontology-engineering Survey and methodological account of ontology construction, ontology uses, and engineering practices. Key claims: - Ontologies support communication, interoperability, and systems engineering. - Ontology construction benefits from explicit scope, competency questions, reuse, and evaluation. - Ontology development should be driven by purpose and intended use. - Competency questions help specify what an ontology must support. - Reuse and evaluation are part of ontology engineering, not optional afterthoughts. ## 37. Ontology Design Patterns for Semantic Web Content - Authors/Org: Aldo Gangemi - Year: 2005 - Venue/Site: International Semantic Web Conference - Bucket: academic - Quality: peer_reviewed - DOI/Identifier: 10.1007/11574620_21 - URL: https://doi.org/10.1007/11574620_21 - Retrieval tags: core, ontology-design-patterns, ontology-engineering, reuse, semantic-web, validation Introduces ontology design patterns as reusable modeling solutions for recurring ontology engineering problems in Semantic Web content. Key claims: - Ontology design patterns improve reuse, interoperability, and modeling quality. - Recurring modeling problems should be solved through shared patterns rather than ad hoc class hierarchies. - LLM-assisted ontology construction can benefit from pattern libraries as constraints and examples. ## 38. ISO/IEC 24707:2018 Information Technology - Common Logic - Authors/Org: International Organization for Standardization and International Electrotechnical Commission - Year: 2018 - Venue/Site: ISO - Bucket: technical - Quality: official_standard - DOI/Identifier: iso/iec 24707:2018 - URL: https://www.iso.org/obp/ui/en - Retrieval tags: common-logic, core, first-order-logic, formal-ontology, interoperability, iso, knowledge-representation, logic, standard International standard for a family of first-order logic-based languages intended for information exchange and transmission of logical content. Key claims: - Common Logic standardizes exchange of logic-based information. - It supports expressive axiomatization beyond lightweight taxonomy or graph structures. - Logic interchange requires precise syntax and semantics. - Common Logic supplies a standard interchange framework for logic-based knowledge representation. - Ontology interchange requires attention to syntax, semantics, and expressivity boundaries. ## 39. LightRAG: Simple and Fast Retrieval-Augmented Generation - Authors/Org: Zirui Guo; et al. - Year: 2024 - Venue/Site: arXiv - Bucket: academic - Quality: secondary_source - DOI/Identifier: arxiv:2410.05779 - URL: https://arxiv.org/abs/2410.05779 - Retrieval tags: core, efficient-rag, graph-rag, graphrag, incremental-update, knowledge-graph, lightrag, llm, local-global-retrieval, rag Proposes a lightweight graph-based RAG approach that combines local and global retrieval over graph structures for efficient generation. Key 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. - Dual-level retrieval can combine entity-level detail with higher-level knowledge discovery. ## 40. Neural-Symbolic Learning and Reasoning: A Survey and Interpretation - Authors/Org: Tarek R. Besold; Artur d'Avila Garcez; Sebastian Bader; Howard Bowman; Pedro Domingos; Pascal Hitzler; Kai-Uwe Kuhnberger; Luis C. Lamb; Daniel Lowd; Priscila Machado Vieira Lima; Leo de Penning; Gadi Pinkas; Hoifung Poon; Gerson Zaverucha - Year: 2017 - Venue/Site: arXiv - Bucket: academic - Quality: peer_reviewed_survey - DOI/Identifier: arxiv:1711.03902 - URL: https://arxiv.org/abs/1711.03902 - Retrieval tags: core, logic, neuro-symbolic, neuro-symbolic-ai, reasoning, survey Survey and interpretation of neural-symbolic methods that combine learning with logic, representation, and reasoning. Key claims: - Neural-symbolic systems seek to combine robust learning with explicit reasoning. - Key issues include representation, extraction, integration, and explainability. - Neural and symbolic AI offer complementary strengths. - Neural-symbolic integration can support reasoning, learning from examples, and interpretability. - Scaling and representation translation remain major challenges. ## 41. Neurosymbolic AI: The 3rd Wave - Authors/Org: Artur d'Avila Garcez; Luis C. Lamb - Year: 2023 - Venue/Site: Artificial Intelligence - Bucket: academic - Quality: peer_reviewed_survey - DOI/Identifier: 10.1016/j.artint.2023.103960 - URL: https://doi.org/10.1016/j.artint.2023.103960 - Retrieval tags: ai, core, neuro-symbolic, reasoning Position paper arguing for neurosymbolic AI as a third wave that integrates neural learning and symbolic reasoning. Key claims: - Neurosymbolic AI aims to combine learning, reasoning, and explanation. - Future AI systems need integration across statistical and symbolic representations. ## 42. Ontology Learning from Text: A Look Back and into the Future - Authors/Org: Wilson Wong; Wei Liu; Mohammed Bennamoun - Year: 2012 - Venue/Site: ACM Computing Surveys - Bucket: academic - Quality: peer_reviewed_survey - DOI/Identifier: 10.1145/2333112.2333115 - URL: https://dl.acm.org/doi/10.1145/2333112.2333115 - Retrieval tags: core, evaluation, ontology-learning, survey, text-mining Survey of ontology learning from text, covering term extraction, concept formation, relation extraction, taxonomy induction, and evaluation. Key claims: - Ontology learning automates parts of ontology construction from textual corpora. - Key challenges include evaluation, relation extraction, and moving from terms to conceptual structure. - Ontology learning remains difficult because semantic interpretation and evaluation are hard. - Different ontology components require different extraction methods. - Future systems need better integration of statistical, linguistic, and knowledge-based methods. ## 43. Ontology Matching: State of the Art and Future Challenges - Authors/Org: Pavel Shvaiko; Jerome Euzenat - Year: 2013 - Venue/Site: IEEE Transactions on Knowledge and Data Engineering - Bucket: academic - Quality: peer_reviewed_survey - DOI/Identifier: 10.1109/tkde.2011.253 - URL: https://doi.org/10.1109/TKDE.2011.253 - Retrieval tags: alignment, core, evaluation, interoperability, ontology-matching, survey Survey of ontology matching approaches, evaluations, systems, and future challenges. Key claims: - Ontology matching combines terminological, structural, semantic, and instance-based evidence. - Open challenges include scalability, uncertainty, user interaction, and evaluation. ## 44. A Translation Approach to Portable Ontology Specifications - Authors/Org: Thomas R. Gruber - Year: 1993 - Venue/Site: Knowledge Acquisition - Bucket: academic - Quality: peer_reviewed_seminal - DOI/Identifier: 10.1006/knac.1993.1008 - URL: https://doi.org/10.1006/knac.1993.1008 - Retrieval tags: conceptualization, core, formal-ontology, gruber, knowledge-sharing, ontology-definition, portability Foundational paper defining an ontology as an explicit specification of a conceptualization and framing ontologies as reusable knowledge-sharing artifacts. Key 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. - Portability requires separating conceptual commitments from implementation details. ## 45. From human experts to machines: An LLM supported approach to ontology and knowledge graph construction - Authors/Org: Vamsi Krishna Kommineni; Birgitta Konig-Ries; Sheeba Samuel - Year: 2024 - Venue/Site: arXiv - Bucket: academic - Quality: preprint - DOI/Identifier: arxiv:2403.08345 - URL: https://arxiv.org/abs/2403.08345 - Retrieval tags: competency-questions, core, evaluation, human-in-the-loop, knowledge-graph-construction, llm, ontology-construction, ontology-engineering Presents a semi-automatic pipeline using LLMs to formulate competency questions, develop an ontology TBox, construct a KG, and evaluate results with reduced human involvement. Key claims: - Competency questions can be used as an organizing bridge between human requirements and LLM-supported ontology construction. - LLM-generated KGs still need evaluation and human-in-the-loop quality checks. - A pipeline view is more useful than single-prompt ontology generation for practical knowledge engineering. - LLMs can reduce expert effort across competency question creation, ontology modeling, and KG construction. - LLM-as-judge evaluation is useful for iteration but cannot replace expert validation. ## 46. Gene Ontology: Tool for the Unification of Biology - Authors/Org: The Gene Ontology Consortium - Year: 2000 - Venue/Site: Nature Genetics - Bucket: academic - Quality: peer_reviewed_seminal - DOI/Identifier: 10.1038/75556 - URL: https://doi.org/10.1038/75556 - Retrieval tags: annotation, biology, biomedical-ontology, controlled-vocabulary, core, data-integration, gene-ontology, scientific-infrastructure Introduces the Gene Ontology as a controlled vocabulary for molecular function, biological process, and cellular component across organisms. Key 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. - Community curation and standard terms are central to scientific data integration. ## 47. Knowledge Representation and Reasoning - Authors/Org: Ronald J. Brachman; Hector J. Levesque - Year: 2004 - Venue/Site: Morgan Kaufmann - Bucket: books - Quality: widely_cited - DOI/Identifier: isbn 9781558609327 - URL: https://shop.elsevier.com/books/knowledge-representation-and-reasoning/brachman/978-1-55860-932-7 - Retrieval tags: book, core, expressivity, knowledge-representation, reasoning, symbolic-ai Textbook covering logic, frames, semantic networks, description logics, reasoning, and KR principles. Key claims: - Knowledge representation is the study of how to encode information so systems can reason with it. - Different KR formalisms make different tradeoffs between expressiveness and reasoning cost. - Representation choices determine what can be inferred and how efficiently. - Knowledge representation must balance expressivity, tractability, and usefulness. - Reasoning systems require formal semantics, not just data structures. ## 48. Model Context Protocol Specification - Authors/Org: Model Context Protocol project - Year: 2025 - Venue/Site: modelcontextprotocol.io - Bucket: technical - Quality: official_docs - DOI/Identifier: mcp specification 2025-06-18 - URL: https://modelcontextprotocol.io/specification/2025-06-18 - Retrieval tags: agents, context-protocol, core, interoperability, mcp, ontology-mcp, semantic-layer, tool-use Official MCP specification defining a protocol for connecting LLM applications with external data sources and tools using standardized client-server interactions. Key claims: - MCP standardizes how LLM applications integrate with external context, resources, and tools. - Tool exposure requires schemas and protocol-level affordances that agents can discover and invoke. - MCP turns tool access into an interoperability layer for agentic AI systems. - Agent systems need a standard interface for tools and contextual resources. - Resources and tools can expose structured enterprise context rather than relying only on text prompts. ## 49. Ontology Development 101: A Guide to Creating Your First Ontology - Authors/Org: Natalya F. Noy; Deborah L. McGuinness - Year: 2001 - Venue/Site: Stanford Knowledge Systems Laboratory - Bucket: technical - Quality: widely_cited - DOI/Identifier: stanford ksl technical report ksl-01-05 - URL: https://protege.stanford.edu/publications/ontology_development/ontology101.pdf - Retrieval tags: classes, core, ontology-development, ontology-engineering, properties, protege, slots, tutorial Practical guide to defining classes, slots, facets, instances, and iterative ontology development using Protege-style modeling. Key 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. - Classes, properties, constraints, and instances should be modeled iteratively. ## 50. Ontology Learning and Population from Text: Algorithms, Evaluation and Applications - Authors/Org: Philipp Cimiano - Year: 2006 - Venue/Site: Springer - Bucket: books - Quality: widely_cited - DOI/Identifier: 10.1007/978-0-387-39252-3 - URL: https://link.springer.com/book/10.1007/978-0-387-39252-3 - Retrieval tags: book, core, nlp, ontology-learning, ontology-population, population, text, text-mining Book on algorithms and evaluation for learning and populating ontologies from text. Key claims: - Ontology learning includes term, synonym, concept, taxonomy, relation, and axiom acquisition. - Text-derived ontologies require evaluation against both linguistic evidence and conceptual adequacy. - Ontology learning includes both schema acquisition and instance population. - Text-derived ontologies need evaluation against expert knowledge and downstream use. - Linguistic and statistical evidence are complementary for concept and relation discovery. ## 51. Retrieval-Augmented Generation of Ontologies from Relational Databases - Authors/Org: Mojtaba Nayyeri; Athish A. Yogi; Nadeen Fathallah; Ratan Bahadur Thapa; Hans-Michael Tautenhahn; Anton Schnurpel; Steffen Staab - Year: 2025 - Venue/Site: arXiv - Bucket: academic - Quality: preprint - DOI/Identifier: arxiv:2506.01232 - URL: https://arxiv.org/abs/2506.01232 - Retrieval tags: core, ontology-generation, owl, rag, relational-database, rigor, schema, semantic-layer, validation Presents RIGOR, a retrieval-augmented iterative approach for generating OWL ontologies from relational database schemas, documentation, domain ontologies, and a growing core ontology. Key claims: - RAG can support ontology generation by retrieving schema documentation, domain ontologies, and prior ontology fragments. - Iterative delta ontology generation with judge-LLM refinement is more controllable than one-shot ontology generation. - Foreign-key structure can guide the order of ontology construction from relational databases. - Relational schemas can be converted into richer OWL ontologies through iterative table-by-table RAG workflows. - Foreign-key structure provides a useful construction order for ontology generation from databases. ## 52. A Review of Relational Machine Learning for Knowledge Graphs - Authors/Org: Maximilian Nickel; Kevin Murphy; Volker Tresp; Evgeniy Gabrilovich - Year: 2016 - Venue/Site: Proceedings of the IEEE - Bucket: academic - Quality: peer_reviewed_survey - DOI/Identifier: 10.1109/jproc.2015.2483592 - URL: https://ieeexplore.ieee.org/document/7358050 - Retrieval tags: core, embedding, embeddings, kg-completion, knowledge-graph, relational-learning Review of statistical relational learning and representation learning methods for knowledge graphs. Key claims: - Knowledge graph embeddings support prediction over relational data. - Relational learning complements symbolic KG semantics with statistical generalization. - Relational structure is central to many AI tasks involving entities and links. - Latent representations can infer missing relations in KGs. - Combining symbolic graph structure with statistical learning is a recurring AI theme. ## 53. KAG: Boosting LLMs in Professional Domains via Knowledge Augmented Generation - Authors/Org: Mingyang Liang; et al. - Year: 2024 - Venue/Site: arXiv - Bucket: academic - Quality: secondary_source - DOI/Identifier: arxiv:2409.13731 - URL: https://arxiv.org/abs/2409.13731 - Retrieval tags: core, enterprise-ai, graph-retrieval, graphrag, hybrid-retrieval, kag, knowledge-augmented-generation, knowledge-graph, llm, logical-form Describes knowledge-augmented generation for professional domains, emphasizing structured knowledge and reasoning over domain-specific sources. Key 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. - Mutual indexing between knowledge graphs and text chunks supports both symbolic reasoning and evidence grounding. ## 54. Ontology Matching - Authors/Org: Jerome Euzenat; Pavel Shvaiko - Year: 2007 - Venue/Site: Springer - Bucket: books - Quality: scholarly_book - DOI/Identifier: 10.1007/978-3-642-38721-0 - URL: https://link.springer.com/book/10.1007/978-3-642-38721-0 - Retrieval tags: alignment, book, core, interoperability, ontology-matching Comprehensive treatment of ontology matching, alignment methods, evaluation, and applications. Key claims: - Ontology matching identifies correspondences between semantically related entities across ontologies. - Alignment is necessary for interoperability across independently built models. ## 55. Agent-OM: Leveraging LLM Agents for Ontology Matching - Authors/Org: Zhangcheng Qiang; Weiqing Wang; Kerry Taylor - Year: 2023 - Venue/Site: arXiv / PVLDB version available - Bucket: academic - Quality: peer_reviewed - DOI/Identifier: arxiv:2312.00326 - URL: https://arxiv.org/abs/2312.00326 - Retrieval tags: agent-memory, core, llm-agent, llm-agents, oaei, ontology-matching, schema-alignment, semantic-interoperability, tool-use, vldb-2025 Introduces an agent-powered ontology matching framework with Siamese agents, retrieval, matching, and memory/tool components, evaluated on OAEI tracks. Key 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. - Agentic matching improves complex and few-shot tasks compared with simpler LLM prompting. ## 56. AIP Evals: Evaluation functions and Ontology edits - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://www.palantir.com/docs/foundry/aip-evals/ontology-edits - Retrieval tags: aip-evals, governance, ontology-edits, palantir, simulation, testing Documentation on evaluating AIP Logic functions that create, edit, or delete ontology objects, including use of Ontology simulations so real ontology state remains unchanged during tests. Key claims: - Logic functions that involve ontology edits are executed in an Ontology simulation during evaluation. - Created objects exist only in the simulated Ontology during tests. - Custom TypeScript evaluation functions can verify edit outcomes. ## 57. AIP observability - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://www.palantir.com/docs/foundry/aip/aip-observability - Retrieval tags: aip-observability, governance, logs, metrics, palantir, tracing Official documentation describing observability for AIP and Ontology workflow executions, including metrics, execution history, distributed tracing, logs, and log search. Key claims: - AIP observability covers AIP and Ontology workflow executions. - Capabilities include metrics, execution history, distributed tracing, logging, and log search. - The feature is part of platform-wide observability. - The feature is part of platform-wide operational observability. ## 58. Cyc: A Midterm Report - Authors/Org: Douglas B. Lenat - Year: 1995 - Venue/Site: Communications of the ACM - Bucket: academic - Quality: peer_reviewed_seminal - DOI/Identifier: 10.1609/aimag.v11i3.842 - URL: https://doi.org/10.1609/aimag.v11i3.842 - Retrieval tags: common-sense, commonsense, core, cyc, knowledge-base, microtheories, symbolic-ai Reports on the Cyc project, a long-running attempt to encode common-sense knowledge in a large symbolic knowledge base. Key claims: - Large-scale common-sense knowledge requires extensive explicit representation. - Symbolic knowledge bases can support reasoning but face acquisition and maintenance challenges. - Common-sense AI requires large amounts of explicit background knowledge. - Hand-built knowledge bases face scale, representation, and inference challenges. - Microtheories can help organize context-dependent knowledge. ## 59. LLMs4OL: Large Language Models for Ontology Learning - Authors/Org: Hamed Babaei Giglou; Jennifer D'Souza; Sören Auer - Year: 2023 - Venue/Site: arXiv - Bucket: academic - Quality: peer_reviewed - DOI/Identifier: arxiv:2307.16648 - URL: https://arxiv.org/abs/2307.16648 - Retrieval tags: benchmark, core, iswc-2023, llm, llms4ol, ontology-engineering, ontology-learning, relation-extraction, schema-extraction, taxonomy Explores the use of large language models for ontology learning tasks, including term typing and taxonomy construction. Key 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. - Term typing, taxonomy discovery, and relation extraction provide a practical task decomposition for LLM-assisted ontology learning. ## 60. Towards a Standard Upper Ontology - Authors/Org: Ian Niles; Adam Pease - Year: 2001 - Venue/Site: FOIS / IEEE - Bucket: academic - Quality: peer_reviewed - DOI/Identifier: 10.1145/505168.505170 - URL: https://doi.org/10.1145/505168.505170 - Retrieval tags: core, formal-ontology, interoperability, standardization, sumo, upper-ontology Presents SUMO as a proposed upper ontology to support interoperability and general-purpose knowledge representation. Key claims: - A standard upper ontology can provide shared high-level categories across systems. - SUMO aims to support broad knowledge engineering reuse. - A standard upper ontology can provide shared high-level concepts for domain ontologies. - SUMO integrates existing ontological resources and formal axiomatization. - Upper ontology can improve semantic interoperability among systems. ## 61. WonderWeb Deliverable D18: Ontology Library (final) - Authors/Org: Claudio Masolo; Stefano Borgo; Aldo Gangemi; Nicola Guarino; Alessandro Oltramari - Year: 2003 - Venue/Site: Laboratory for Applied Ontology - Bucket: academic - Quality: widely_cited - DOI/Identifier: wonderweb d18 - URL: http://www.loa.istc.cnr.it/old/DOLCE.html - Retrieval tags: core, dolce, foundational-ontology, upper-ontology Introduces DOLCE and related foundational ontologies for cognitive and linguistic engineering. Key claims: - DOLCE models categories such as endurants, perdurants, qualities, and social objects. - Foundational ontologies can clarify domain modeling choices. ## 62. Palantir Technologies Inc. Form 10-K for fiscal year ended December 31, 2025 - Authors/Org: Palantir Technologies Inc. - Year: 2026 - Venue/Site: U.S. Securities and Exchange Commission EDGAR - Bucket: palantir - Quality: primary_source - DOI/Identifier: sec accession 0001321655-26-000011 - URL: https://www.sec.gov/Archives/edgar/data/1321655/000132165526000011/pltr-20251231.htm - Retrieval tags: 10-k, aip, foundry, investor, ontology, palantir, revenue, risk-factors, sec Annual report filed February 17, 2026. Describes Palantir's four principal platforms, including Foundry, AIP, Apollo, and Gotham; provides customer, revenue, risk, and business-strategy disclosures. Key claims: - Foundry provides data management, logic authoring, systemic mapping through Ontology, analytics, and workflow development. - AIP provides secure LLM connectivity, agent/automation development tooling, AI-enabled applications, and evaluation frameworks. - Palantir reported 954 customers as of December 31, 2025 and total 2025 revenue of about $4.475 billion. - Palantir describes its platforms and business strategy in primary investor-facing terms. - The filing is useful for distinguishing product architecture claims from business growth and risk disclosures. ## 63. Artificial Intelligence: A Modern Approach - Authors/Org: Stuart Russell; Peter Norvig - Year: 2020 - Venue/Site: Pearson - Bucket: books - Quality: scholarly_book - DOI/Identifier: isbn 9780134610993 - URL: https://aima.cs.berkeley.edu - Retrieval tags: agents, ai, book, core, knowledge-representation Standard AI textbook covering agents, search, logic, probabilistic reasoning, learning, natural language, and robotics. Key claims: - AI can be organized around rational agents that perceive, reason, learn, and act. - Knowledge representation and reasoning are core components of intelligent systems. ## 64. AI ethics and governance - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/aip/ethics-governance - Retrieval tags: ai-governance, aip-evals, aip-logic, data-health, debugging, evals, interpretable-tools, palantir, responsible-ai Official guidance on AI ethics and governance in AIP, emphasizing interpretable tool delegation, debug visibility, testing, evaluation, and operational governance. Key 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. - Experiments can assess how input changes affect model responses. ## 65. AIP observability: Service logs and debugging - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/aip-observability/service-logs-and-debugging - Retrieval tags: aip-observability, debugging, logs, palantir, permissions, privacy Official page explaining log access and debugging permissions for traces and service logs in AIP observability. Key claims: - Administrators must enable log access for relevant projects to view traces and service logs. - Users generally have access to logs for their own executions from the past 24 hours, with stack-specific caveats. - Log access is permissioned rather than universally visible. ## 66. AIP security and privacy - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/aip/aip-security - Retrieval tags: aip-security, bias, governance, human-judgment, palantir, privacy Official documentation and FAQ describing Palantir's AIP security and privacy posture, including customer data protection, responsible processing, privacy, security, bias, and human judgment concerns. Key claims: - Palantir frames privacy and security as first principles for AIP deployment. - The documentation acknowledges generative AI concerns around privacy, security, bias, discrimination, and human judgment. - Security and privacy claims must be distinguished from independent audit evidence. ## 67. AIP Virtual Tables - Authors/Org: Palantir Technologies - Year: 2024 - Venue/Site: Palantir Blog - Bucket: palantir - Quality: official_docs - URL: https://blog.palantir.com/aip-virtual-tables-5094b5e4b3bd - Retrieval tags: aip, architecture, data-integration, palantir, virtual-tables Palantir blog describing virtual tables as a way to expose external data to AIP and ontology workflows without always physically copying data. ## 68. Building with Palantir AIP: Logic Tools for RAG/OAG - Authors/Org: Palantir Technologies - Year: 2024 - Venue/Site: Palantir Blog - Bucket: palantir - Quality: official_docs - URL: https://blog.palantir.com/building-with-palantir-aip-logic-tools-for-rag-oag-fdaf8938d02e - Retrieval tags: aip, logic-tools, oag, ontology, palantir, rag Palantir technical blog describing logic tools for retrieval-augmented and ontology-augmented generation, relevant to how AIP connects LLMs to governed operational logic. ## 69. Data security overview - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://www.palantir.com/docs/foundry/security/data-security-overview - Retrieval tags: data-security, foundry-security, governance, palantir, permissions Official Foundry security overview covering data access, permissions, and governance concepts that underpin ontology-backed workflows. Key claims: - Foundry security controls are designed to govern access to data and derived resources. - Permissioning is a platform-level concern across data, applications, and ontology resources. - Enterprise AI workflows depend on these controls when exposing data to tools or models. ## 70. Foundry platform summary for LLMs - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://www.palantir.com/docs/foundry/getting-started/foundry-platform-summary-llm - Retrieval tags: claude, copilot-studio, external-agents, foundry, gemini-enterprise, llm, mcp-hub, oauth, ontology-mcp, palantir Official Palantir documentation summarizing Foundry capabilities for LLMs, including Ontology MCP, AIP, tool use, permissions, and external agent integrations. Key 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. - MCP Hub provides a central location to discover and manage Ontology MCP servers configured across an enrollment. ## 71. Functions: Overview - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/functions/overview - Retrieval tags: architecture, foundry, functions, logic, ontology, palantir Official overview of Foundry Functions, which expose custom code and logic into applications, ontology workflows, and operational decision processes. ## 72. Manage published functions - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/functions/manage-functions - Retrieval tags: functions, governance, lifecycle, palantir, published-functions Official documentation for managing published functions, relevant to lifecycle and operational governance of reusable logic attached to ontology workflows. ## 73. Object security: Overview - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://www.palantir.com/docs/foundry/security/object-security/overview - Retrieval tags: access-control, foundry, governance, object-security, palantir, permissions Official documentation for object security in Foundry, describing how ontology object access can be governed through security controls and policies. Key claims: - Foundry supports security controls over ontology object access. - Object security is a platform-level governance mechanism for application and workflow access. - Security rules can shape what users and tools are allowed to see or use. ## 74. Ontology Manager: Overview - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/ontology-manager/overview - Retrieval tags: action-types, object-types, ontology-lifecycle, ontology-manager, palantir Official documentation for Ontology Manager, the application used to build and maintain object types, action types, data connections, and ontology health. Key claims: - Ontology Manager supports creation and maintenance of object types and action types. - It connects data to the Ontology and helps investigate updating behavior in user applications. - Ontology management is treated as an ongoing operational process. ## 75. Ontology: Core concepts - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://www.palantir.com/docs/foundry/ontology/core-concepts - Retrieval tags: action-type, link-type, object-type, ontology, palantir, property, role, roles Defines core ontology concepts: object types, objects, properties, link types, action types, and roles, with a comparison to datasets, rows, columns, and joins. Key claims: - An ontology maps datasets and models to object types, properties, link types, and action types. - Object types represent real-world entities or events. - Action types define changes or edits that users can apply. - Properties correspond to object attributes, while link types represent relationships between object types. - Action types define controlled changes or edits users can apply to objects, properties, and links. ## 76. Palantir MCP: Security - Data governance - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/palantir-mcp/security - Retrieval tags: ai-systems, data-governance, foundry-resources, mcp-security, palantir Official Palantir MCP security page describing data governance and data-flow/security models for MCP integration between Foundry and AI systems. Key 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. ## 77. Building with Palantir AIP: Data Tools for RAG / OAG - Authors/Org: Palantir Technologies - Year: 2023 - Venue/Site: Palantir Blog - Bucket: palantir - Quality: primary_source - URL: https://blog.palantir.com/building-with-palantir-aip-data-tools-for-rag-oag-b3b509c8b0f3 - Retrieval tags: aip, llm, llm-grounding, oag, ontology-augmented-generation, palantir, palantir-blog, rag Vendor-authored blog introducing Ontology Augmented Generation as a broader decision-centric extension of retrieval augmented generation using ontology data and relationships. Key claims: - OAG is presented as a decision-centric version of RAG. - LLMs can retrieve context-specific business data such as orders, customers, and locations. - Ontology context is framed as a way to reduce hallucination and improve relevance. - Palantir distinguishes document retrieval from ontology-aware retrieval and tool use. - Ontology Augmented Generation frames objects, relationships, and actions as AI context. ## 78. Privacy and Civil Liberties - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir - Bucket: palantir - Quality: primary_source - URL: https://www.palantir.com/pcl - Retrieval tags: civil-liberties, governance, palantir, privacy, public-sector, public-trust Official Palantir page describing its privacy and civil liberties posture, including claims about preserving privacy and civil liberties while using data. Key claims: - Palantir positions privacy and civil liberties as part of its founding mission. - The company claims its software supports controlled data use rather than selling customer data. - The page is part of Palantir's response to surveillance and public trust concerns. - Palantir presents privacy and civil liberties as part of its corporate operating model. - The page emphasizes governance, legal compliance, and responsible use themes. ## 79. Action types - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://www.palantir.com/docs/foundry/ontology/action-types - Retrieval tags: action-types, governance, ontology, operational-actions, palantir, writeback Official documentation for action types, the Ontology primitive used to define sanctioned changes to objects, properties, links, and related operational side effects. Key claims: - Action types define controlled modifications to ontology objects and links. - Action definitions can include parameters, validation, and operational side effects. - Action types are the primary public primitive for turning an ontology from a read model into an operational interface. ## 80. AIP Analyst: Overview - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://www.palantir.com/docs/foundry/aip-analyst/overview - Retrieval tags: agentic-analysis, aip-analyst, object-sets, ontology-search, palantir Documentation for AIP Analyst, an agentic interface that answers natural-language questions by searching ontology resources, creating object sets, transforming data, and generating outputs. Key claims: - AIP Analyst searches across the ontology for relevant data. - It can create object sets and perform transformations before summarizing or visualizing results. - It uses tools such as object type search, object type lookup, object search, and dataset lookup. ## 81. AIP architecture overview - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Architecture Center - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/architecture-center/aip-architecture - Retrieval tags: agentic-ai, aip-architecture, governance, logging, observability, palantir Official architecture page describing observability, logging for human and AI actions, ontology data flows, chained executions, and token/resource usage monitoring. Key claims: - AIP architecture includes monitoring for AI-driven workflows and agentic processes. - The page describes logging for actions taken by human users or AI agents. - Observability includes tracing cascades of chained executions and token/resource usage. ## 82. AIP Logic compute usage - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://www.palantir.com/docs/foundry/logic/compute-usage - Retrieval tags: aip-logic, compute-usage, governance, palantir, resource-management Official documentation for understanding and managing compute use in AIP Logic workflows. Key claims: - AIP Logic has measurable compute usage tied to workflow execution. - Resource management is part of the production AI workflow story. - Cost and capacity governance can be treated as operational controls, not only billing afterthoughts. ## 83. AIP observability: Tracing - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/aip-observability/trace-view - Retrieval tags: aip-observability, distributed-tracing, palantir, security-boundaries, workflow-execution Official trace-view documentation explaining distributed traces across services, networks, process boundaries, and security boundaries. Key claims: - Trace view shows a visual timeline of workflow execution. - Distributed traces can cross process, network, and security boundaries. - Tracing helps understand the path a request takes through an application. ## 84. Discover and manage Ontology MCP servers in MCP Hub - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Announcements - Bucket: palantir - Quality: official_docs - URL: https://www.palantir.com/docs/foundry/announcements/2026-05 - Retrieval tags: data-governance, external-agents, mcp, mcp-hub, ontology-mcp, palantir, permissions, phase5, security, tool-use May 2026 Palantir announcement stating that Ontology MCP servers are discoverable through MCP Hub, with central management for MCP servers configured across an enrollment. Key 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. - Palantir warns that enabling Ontology MCP makes ontology resources available to external MCP clients and should be checked against data-governance and security policies. ## 85. Link types - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://www.palantir.com/docs/foundry/ontology/link-types - Retrieval tags: link-types, ontology, operational-graph, palantir, relationships Official documentation for defining link types between object types in the Ontology, including relationship semantics and navigation over connected operational objects. Key claims: - Link types define relationships between object types. - Relationship modeling enables graph-style navigation across operational entities. - Links make the Ontology more than isolated tables by encoding cross-object context. ## 86. Object edits and materializations: Enable user edit history - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/object-edits/user-edit-history - Retrieval tags: audit, edit-history, object-edits, object-storage, palantir, provenance Official documentation on tracking history of user edits to objects indexed into Object Storage V2 and surfacing edit histories in Workshop modules or object views. Key claims: - Edit history is a concrete provenance mechanism for user-driven ontology object changes. - Disabling edit history can permanently delete existing histories, making retention policy a governance issue. - Operational ontology platforms need object-level change histories for auditability and trust. ## 87. Object edits and materializations: Manage schema changes - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/object-edits/schema-migrations - Retrieval tags: object-edits, object-storage, ontology-governance, palantir, schema-migration, validation Official documentation describing breaking and non-breaking schema changes for object types, user edits, and Object Storage V2 schema migration support. Key claims: - Operational ontology schemas evolve, and breaking changes require migration support when user edits exist. - Object Storage V2 provides a schema migration framework for certain ontology object changes. - Schema governance is essential when ontology-backed workflows become writable and persistent. ## 88. Object security: Security controls - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://www.palantir.com/docs/foundry/security/object-security/security-controls - Retrieval tags: governance, object-security, ontology-permissions, palantir, security-controls Official documentation describing object security controls for constraining visibility or access to ontology objects and properties in Foundry. Key claims: - Object security controls can be configured to govern object visibility and access. - Security controls are tied to ontology resources rather than only raw storage locations. - Governance is part of the operational ontology model, not an external afterthought. ## 89. Object types - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://www.palantir.com/docs/foundry/ontology/object-types - Retrieval tags: data-modeling, foundry, object-types, ontology, palantir Official documentation for configuring object types in the Ontology, including properties, display metadata, backing data, and lifecycle considerations. Key claims: - Object types define classes of real-world entities, events, or concepts in Foundry Ontology. - Object type configuration binds operational semantics to backing datasets or other data sources. - Object type design shapes how applications, SDKs, and AI tools see enterprise data. ## 90. Palantir MCP: Installation - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/palantir-mcp/installation - Retrieval tags: boundary, data-governance, external-agents, mcp, palantir, permissions Official installation page stating that Palantir MCP enables external AI systems to read Foundry data/metadata and interact via AI-friendly API endpoints using configured user-token permissions, while warning that external-system governance matters after data leaves Palantir. Key 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. ## 91. Palantir MCP: Overview - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/palantir-mcp/overview - Retrieval tags: ai-ides, foundry, ontology-modification, palantir, palantir-mcp Official overview of Palantir MCP, which provides tools for external AI systems to take actions in Foundry, search or modify the ontology, and update Developer Console applications. Key claims: - Palantir MCP can search the ontology and safely modify the ontology. - It can update Developer Console applications and generate OSDK-related changes. - The toolset supports AI IDEs and external agents interacting with Foundry. ## 92. Platform overview - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://www.palantir.com/docs/foundry/platform-overview/overview - Retrieval tags: ai-mesh, aip, apollo, foundry, ontology-layer, palantir, platform-overview Palantir platform overview describing AIP, Foundry, Apollo, and the ontology layer as an AI Mesh for connecting generative AI to operations. Key claims: - AIP connects generative AI to operations. - Foundry and AIP include an ontology layer plus core services and security/governance layers. - The Ontology is positioned as representing enterprise decisions, not only data. - Foundry and AIP include an ontology layer plus core services and security and governance layers. ## 93. Connecting Agents to Decisions - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Blog - Bucket: palantir - Quality: primary_source - URL: https://blog.palantir.com/connecting-agents-to-decisions-277dee8ddb40 - Retrieval tags: agent-memory, agents, decision-centric, decision-lineage, governed-action, human-agent-workflows, ontology, palantir, phase5, security Palantir-authored technical narrative describing the Ontology as a decision-centric architecture for human-agent workflows, bringing data, logic, action, and security into a single operating model. Key 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. - Decision lineage and telemetry are presented as part of the memory and governance substrate for agents. ## 94. FDP Data Protection Impact Assessment: NDIT Identifiable Version v3.0 - Authors/Org: NHS England - Year: 2026 - Venue/Site: NHS England - Bucket: palantir - Quality: primary_source - DOI/Identifier: redacted-ndit-nhs-england-fdp-dpia-identifiable-version-v3.0 - URL: https://www.england.nhs.uk/wp-content/uploads/2025/08/redacted-ndit-nhs-england-fdp-dpia-identifiable-version-v3.0.pdf - Retrieval tags: data-protection, dpia, fdp, identifiable-data, ndit, nhs, ontology-definition, palantir, phase6, privacy Redacted DPIA for identifiable data processing in the FDP/NDIT context. It defines the FDP federation, Platform Contractor, NHS-PET, purposes, access controls, opt-out concepts, and the Ontology layer. Key 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. ## 95. Ontological foundations for structural conceptual models - Authors/Org: Giancarlo Guizzardi - Year: 2005 - Venue/Site: University of Twente Research Information - Bucket: academic - Quality: scholarly_index - URL: https://research.utwente.nl/en/publications/ontological-foundations-for-structural-conceptual-models - Retrieval tags: commercial, conceptual-modeling, core, model-quality, ontouml, openalex The main objective of this thesis is to contribute to the theory of Conceptual Modeling by proposing ontological foundations for structural conceptual models. Conceptual Modeling is a discipline of great importance to several areas in Computer Science. Its main objective is concerned with identifying, analyzing and describing the essential concepts and constraints of a universe of discourse, with the help of a (diagrammatic) modeling language that is based on a set of basic modeling concepts (forming a metamodel). In this thesis, we show how conceptual modeling languages can be evaluated and (re)designed with the purpose of improving their ontological adequacy. In simple terms, ontological adequacy is a measure of how close the models produced using a modeling language are to the situations in the reality they are supposed to represent. The thesis starts by proposing a systematic evaluation method for comparing a metamodel of the concepts underlying a language to a reference ontology of the corresponding domain in reality. The focus of this thesis is on general conceptual modeling languages (as opposed to domain specific ones). Hence, the proposed reference ontology is a foundational (or upper-level) ontology. Moreover, since, it focuses on structural modeling aspects (as opposed to dynamic ones), this foundational ontology is an ontology of objects, their properties and relations, their parts, the roles they play, and the types they instantiate. The proposed ontology was developed by adapting and extending a number of theories coming, primarily, from formal ontology in philosophy, but also from cognitive science and linguistics. Once developed, every subtheory of the ontology is used in the creation of methodological tools (e.g., modeling profiles, guidelines and design patterns). The expressiveness and relevance of these tools are shown throughout the thesis to solve some classical and recurrent conceptual modeling problems. Finally, the thesis demonstrates the applicability and usefulness of both the method and the proposed ontology by analyzing and extending a fragment of the Unified Modeling Language (UML) which deals with the construction of structural conceptual models. Key claims: - Conceptual models need ontological foundations to avoid category mistakes. - OntoUML and related ideas help distinguish kinds, roles, phases, and relators. ## 96. A Direct Mapping of Relational Data to RDF - Authors/Org: Marcelo Arenas; Alexandre Bertails; Eric Prud'hommeaux; Juan Sequeda; W3C RDB2RDF Working Group - Year: 2012 - Venue/Site: W3C Recommendation - Bucket: technical - Quality: official_standard - URL: https://www.w3.org/TR/rdb-direct-mapping - Retrieval tags: data-integration, direct-mapping, knowledge-graph-construction, mapping-language, phase15, rdf, relational-data, w3c Direct Mapping is the W3C Recommendation for defining a default RDF graph representation of relational database content and schema. It complements R2RML by providing a canonical baseline for relational-to-RDF translation. Key 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. ## 97. Dataspace Protocol 2025-1 - Authors/Org: Eclipse Dataspace Protocol project; International Data Spaces Association contributors - Year: 2025 - Venue/Site: Eclipse Dataspace Protocol Specification - Bucket: technical - Quality: official_standard - URL: https://eclipse-dataspace-protocol-base.github.io/DataspaceProtocol/2025-1 - Retrieval tags: agreement-negotiation, data-space, dataspace-protocol, dataspaces, federated-data-sharing, phase15, usage-control, usage-policy The Dataspace Protocol 2025-1 specification defines schemas and protocols for entities to publish data, negotiate agreements, and access data as part of federated dataspaces. It builds on Web technologies and usage-control-oriented interoperability. Key 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. ## 98. eXtensible Access Control Markup Language (XACML) Version 3.0 - Authors/Org: OASIS eXtensible Access Control Markup Language Technical Committee - Year: 2013 - Venue/Site: OASIS Standard - Bucket: technical - Quality: official_standard - DOI/Identifier: oasis standard, 22 january 2013 - URL: https://docs.oasis-open.org/xacml/3.0/xacml-3.0-core-spec-os-en.html - Retrieval tags: access-control, attribute-based-access-control, governed-action, oasis, ontology-governance, phase16, policy-enforcement, runtime-governance, xacml XACML 3.0 defines a policy language and request/response model for attribute-based access control, including policies, rules, targets, obligations, advice, combining algorithms, and decision points. Key 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. ## 99. IEC Common Data Dictionary (IEC CDD) - Authors/Org: International Electrotechnical Commission - Year: 2025 - Venue/Site: International Electrotechnical Commission - Bucket: technical - Quality: official_standard - DOI/Identifier: iec 61360 / iec common data dictionary - URL: https://cdd.iec.ch - Retrieval tags: common-data-dictionary, data-dictionary, iec, industrial-semantic-standards, ontology-governance, phase16, product-classification, semantic-interoperability IEC CDD is the IEC Common Data Dictionary for electrotechnical product classes and properties. It provides standardized machine-readable product semantics for components, properties, and classification used across industrial and standards ecosystems. Key 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. ## 100. R2RML: RDB to RDF Mapping Language - Authors/Org: Souripriya Das; Seema Sundara; Richard Cyganiak; W3C RDB2RDF Working Group - Year: 2012 - Venue/Site: W3C Recommendation - Bucket: technical - Quality: official_standard - URL: https://www.w3.org/TR/r2rml - Retrieval tags: data-integration, knowledge-graph, knowledge-graph-construction, mapping-language, mappings, phase15, r2rml, rdf, relational-data, w3c R2RML is the W3C Recommendation for expressing customized mappings from relational databases to RDF datasets. It defines logical tables, subject maps, predicate-object maps, joins, and graph maps for generating RDF from relational source data. Key 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. ## 101. RDF 1.2 Concepts and Abstract Data Model - Authors/Org: World Wide Web Consortium - Year: 2026 - Venue/Site: W3C Candidate Recommendation Snapshot - Bucket: technical - Quality: official_standard - DOI/Identifier: w3c cr-rdf12-concepts-20260407 - URL: https://www.w3.org/TR/rdf12-concepts - Retrieval tags: graph-data, rdf-1-2, semantic-web, standard, w3c Current W3C Candidate Recommendation snapshot for RDF 1.2 concepts and abstract data model, updating the RDF graph model used by semantic web systems. Key claims: - RDF remains the foundational graph data model for interoperable semantic data. - Standard abstract graph semantics matter for exchange across tools and organizations. - RDF 1.2 work shows active maintenance of the Semantic Web standards stack. ## 102. SPARQL 1.1 Federated Query - Authors/Org: W3C SPARQL Working Group - Year: 2013 - Venue/Site: W3C Recommendation - Bucket: technical - Quality: official_standard - URL: https://www.w3.org/TR/sparql11-federated-query - Retrieval tags: distributed-query, federated-data-sharing, federated-query, phase16, rdf, semantic-interoperability, sparql, w3c SPARQL 1.1 Federated Query extends SPARQL with SERVICE-based subqueries over remote endpoints. It provides a standards route for querying distributed RDF data without materializing every graph into one store. Key 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. ## 103. AIP Chatbot Studio: Overview - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://www.palantir.com/docs/foundry/chatbot-studio/overview - Retrieval tags: aip-agent-studio, aip-chatbot-studio, llm-agents, palantir, read-write-workflows Documentation for building AIP Chatbots, formerly AIP Agents, powered by LLMs, the Ontology, documents, and custom tools, with internal and external deployment paths. Key claims: - AIP Chatbots are equipped with enterprise-specific information and tools. - They are powered by LLMs, the Ontology, documents, and custom tools. - They can support dynamic, context-aware read and write workflows. ## 104. AIP Evals: Overview - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/aip-evals/overview - Retrieval tags: aip-evals, llm-evaluation, palantir, regression, test-cases Official overview of AIP Evals as a testing environment for AIP Logic, chatbot, and code-authored functions, designed for nondeterministic LLM behavior. Key claims: - AIP Evals supports test cases and evaluation functions. - It compares results against previous versions of a function. - It is designed to address nondeterminism in LLM applications. ## 105. AIP features - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://www.palantir.com/docs/foundry/aip/aip-features - Retrieval tags: agents, aip-features, llm-workflows, ontology-data, palantir, palantir-mcp Official documentation listing AIP application and builder capabilities, including LLM-backed workflows, agents, applications, OSDK, AIP Logic, and Palantir MCP. Key claims: - AIP enables developers to build LLM-backed workflows, agents, and applications. - Developer tooling provides access to ontology data, logic, and actions. - Palantir MCP connects external AI IDEs and agents to ontology and Foundry context. ## 106. AIP Logic metrics - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://www.palantir.com/docs/foundry/logic/aip-logic-metrics - Retrieval tags: aip-logic, metrics, observability, palantir, workflow-monitoring Official documentation for metrics associated with AIP Logic, supporting monitoring of workflow behavior and resource use. Key claims: - AIP Logic exposes metrics that can be used to monitor logic function execution. - Metrics make AI workflow behavior observable beyond prompt outputs. - Execution metrics support governance, performance analysis, and operational troubleshooting. ## 107. AIP Logic: Core concepts - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://www.palantir.com/docs/foundry/logic/core-concepts - Retrieval tags: aip-logic, blocks, palantir, tools, typed-inputs, workflow-composition Official concept page for AIP Logic, covering logic functions, blocks, inputs, outputs, tools, and execution concepts. Key claims: - AIP Logic workflows are composed from blocks with typed inputs and outputs. - LLM calls can be combined with tools and deterministic business logic. - Logic concepts create a bridge between AI generation and governed operational execution. ## 108. AIP Logic: Overview - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/logic/overview - Retrieval tags: aip-logic, human-review, llm-tools, logic-functions, ontology, ontology-edits, palantir, workflow, workflows Official overview of AIP Logic, Palantir's tool for creating logic functions and AI workflows using blocks, models, tools, and ontology-connected inputs and outputs. Key claims: - AIP Logic is a builder surface for AI-powered workflows and logic functions. - Logic workflows can combine LLM calls, tools, ontology data, and deterministic processing. - Logic functions can be reused by applications, agents, and other Foundry components. - AIP Logic functions can be automated. - Ontology edits can be automatically applied or staged for human review. ## 109. AIP Model Catalog: Overview - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/model-catalog/overview - Retrieval tags: aip, architecture, model-catalog, model-governance, palantir Official documentation for Model Catalog, the Foundry service used to register and manage AI models available to AIP workflows. ## 110. AIP observability: Execution history - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/aip-observability/run-history - Retrieval tags: actions, auditability, automations, functions, palantir, run-history Official documentation for run history of functions, actions, and automations, including recent executions in a resource-level run-history tab. Key claims: - Run history is available for a Function, Action, or automation. - The run-history tab shows recent executions. - Execution history gives a resource-level view of operational workflow runs. ## 111. Functions in the Ontology - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://www.palantir.com/docs/foundry/ontology/functions - Retrieval tags: business-logic, functions, ontology, palantir, tools Official documentation describing functions as logic attached to the Ontology and usable by applications, object views, SDKs, and workflows. Key claims: - Functions define reusable logic over ontology resources. - Functions can extend ontology behavior beyond stored properties and links. - Functions provide a controlled logic surface for applications and AI-enabled workflows. ## 112. Functions on objects (FOO) - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/functions/functions-on-objects - Retrieval tags: functions, lineage, object-sets, ontology-objects, palantir, transparency Official documentation describing functions that operate natively on ontology objects and object sets, including retrieval, storage, modification, lineage, and transparency claims. Key claims: - Functions can take object and object-set types as parameters. - Functions can search object sets and modify objects using OntologyEditFunctions. - Palantir states functions go beyond FaaS by adding native ontology support, data security, lineage, and transparency. ## 113. Migrate to project-based permissions - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/ontology-manager/migrate-to-project-based-permissions - Retrieval tags: compass, governance, ontology-permissions, palantir, project-permissions Official page explaining project-based permissions for ontology resources such as object types, action types, link types, interfaces, and shared properties. Key claims: - Ontology resources can be saved within projects and inherit project permissions. - Object and link instance permissions remain dependent on backing datasource location. - Permissions to view, edit, and manage ontology resources are administered through Compass. ## 114. Ontology MCP: Authentication and authorization - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/ontology-mcp/authentication-and-authorization - Retrieval tags: governed-action, mcp, oauth, ontology-mcp, palantir, permissions, security Official documentation explaining how Ontology MCP uses the OAuth 2.0 configuration, application restrictions, permissions, and scoped tokens of the Developer Console application. Key claims: - Ontology MCP uses existing OAuth 2.0 application configuration rather than a separate authentication system. - Application restrictions and permissions apply to MCP requests made by external agents. - Scoped tokens define which ontology operations and resources an MCP client can access. ## 115. Ontology MCP: Getting started - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://www.palantir.com/docs/foundry/ontology-mcp/getting-started - Retrieval tags: agent-configuration, developer-console, external-agents, mcp-hub, ontology-mcp, palantir, phase5, tool-use Official getting-started guide for Ontology MCP, including discovery through MCP Hub and configuration of MCP servers for ontology resources. Key claims: - Ontology MCP servers are discoverable through the MCP Hub application. - MCP Hub lists MCP servers configured on an enrollment and links to their Developer Console applications. - Ontology MCP server details include the tools and ontology resources exposed to agents. - The guide positions MCP Hub as a workflow for managing external agent access to ontology resources. ## 116. Ontology MCP: MCP tools and agent configuration - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/ontology-mcp/mcp-tools-and-agent-configuration - Retrieval tags: actions, agent-configuration, ontology-mcp, palantir, tool-description Official documentation explaining how ontology actions can be described for AI agents through tool descriptions and configuration fields. Key claims: - Ontology Manager includes an Agent tool description field for actions exposed to agents. - Tool descriptions guide AI agents on when and how to use each action. - Action configuration becomes part of the agent-facing contract. ## 117. Ontology MCP: Sample architecture - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://www.palantir.com/docs/foundry/ontology-mcp/sample-architecture - Retrieval tags: agents, controlled-actions, mcp, ontology-mcp, palantir, tool-use Documentation describing Ontology MCP as a way to expose object types, action types, and query functions from a Developer Console application as tools for agents. Key claims: - Ontology MCP exposes object types, action types, and query functions to agents. - Object types can be reachable through SQL tools. - Action types can become individual MCP tools for controlled writes. ## 118. Ontology SDK: TypeScript OSDK - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://palantir.com/docs/foundry/ontology-sdk/typescript-osdk - Retrieval tags: actions, functions, ontology, ontology-sdk, osdk, palantir, queries, typed-client, typescript, typescript-osdk Documentation for generated TypeScript SDKs that expose ontology object types, actions, and functions to application developers. Key claims: - Action types in the Ontology become predefined operations in generated SDK code. - The OSDK exposes objects, actions, and functions for application development. - Functions can define custom logic on ontology data. - Generated SDK code exposes object types, action types, and functions from the Ontology. - Action types become predefined operations in application code. ## 119. Permission checks when applying an Action - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://www.palantir.com/docs/foundry/object-edits/permission-checks - Retrieval tags: access-control, action-permissions, ontology-security, palantir, submission-criteria Official documentation describing permission checks for Actions, including different behavior for single-datasource and multi-datasource object types and submission criteria. Key claims: - Action permission checks depend on object backing architecture. - For single-datasource edits, users need object view/load access and must pass action submission criteria. - Creating new objects requires access to the input datasource. ## 120. Why create an Ontology? - Authors/Org: Palantir Technologies - Year: 2026 - Venue/Site: Palantir Foundry Documentation - Bucket: palantir - Quality: official_docs - URL: https://www.palantir.com/docs/foundry/ontology/why-ontology - Retrieval tags: agents, data-logic-action-security, decision-centric, operations, palantir Explains the value proposition for an ontology in Foundry: decision-centric modeling with data, logic, action, and security as the elements of operational decisions. Key claims: - Palantir models operational decisions as data, logic, action, and security. - The decision-centric Ontology connects humans and agents to operations. - The Ontology is framed as more than data cataloging or schema design. - The Ontology connects humans and agents to operational decisions. - The Ontology is framed as more than cataloging data or designing schemas. ## 121. AllegroGraph knowledge graph, vector, and LLM materials - Authors/Org: Franz Inc. - Year: 2026 - Venue/Site: AllegroGraph / Franz Inc. - Bucket: technical - Quality: official_docs - URL: https://allegrograph.com - Retrieval tags: allegrograph, graphrag, llm, neuro-symbolic, rdf, vector-search AllegroGraph materials position RDF knowledge graphs, reasoning, vector storage, and LLM/RAG capabilities as a combined neuro-symbolic AI platform. Key claims: - RDF graph storage can be combined with vector similarity and LLM workflows. - Reasoning and semantic relationships can provide explainable structure for AI context. - Neuro-symbolic positioning joins curated symbols with statistical retrieval. ## 122. Connecting AI to Decisions with the Palantir Ontology - Authors/Org: Akshay Krishnaswamy, Palantir - Year: 2024 - Venue/Site: Palantir Blog - Bucket: palantir - Quality: primary_source - URL: https://blog.palantir.com/connecting-ai-to-decisions-with-the-palantir-ontology-c73f7b0a1a72 - Retrieval tags: aip, blog, decision-centric-ontology, decisions, enterprise-decisions, ontology, operational-ai, palantir, palantir-blog Vendor-authored blog explaining Palantir's decision-centric ontology thesis and why Palantir argues operational AI depends on modeling decisions, actions, and context rather than only data. Key claims: - The Ontology is designed to represent decisions in an enterprise, not simply data. - Traditional data architectures do not capture reasoning or action well enough for operational AI. - AIP's operational impact is attributed to software architecture around the Ontology. - Palantir argues that AI must be connected to operational decisions rather than isolated chat interfaces. - The Ontology is presented as the structure that maps enterprise data, logic, action, and security to decision workflows. ## 123. Data Privacy Vocabulary (DPV) Version 2.0 - Authors/Org: W3C Data Privacy Vocabularies and Controls Community Group - Year: 2024 - Venue/Site: W3C Community Final Specification - Bucket: technical - Quality: official_docs - URL: https://www.w3.org/community/reports/dpvcg/CG-FINAL-dpv-20240801 - Retrieval tags: compliance, data-governance, dpv, gdpr, ontology-governance, phase16, policy-vocabulary, privacy, privacy-vocabulary, risk The Data Privacy Vocabulary 2.0 provides a machine-readable vocabulary for personal data handling, purposes, processing operations, legal bases, rights, risks, controls, and organizational measures. Key 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. ## 124. eccenca Corporate Memory knowledge graph and data product governance materials - Authors/Org: eccenca - Year: 2026 - Venue/Site: eccenca - Bucket: commercial - Quality: official_docs - URL: https://eccenca.com/products/corporate-memory - Retrieval tags: ai-ready-kg, corporate-memory, data-products, eccenca, semantic-governance, shacl eccenca Corporate Memory materials cover knowledge graph lifecycle management, semantic data governance, SHACL validation, metadata, and AI-ready semantic context. Key claims: - Knowledge graphs require lifecycle workflows for modeling, validating, publishing, and reuse. - Semantic data products need provenance, stewardship, and quality controls. - SHACL and governance workflows can make ontology rules enforceable. ## 125. eCl@ss in RDF/OWL - Authors/Org: eCl@ss e.V. - Year: 2025 - Venue/Site: eCl@ss Standard Documentation - Bucket: technical - Quality: official_docs - DOI/Identifier: ecl@ss rdf/owl distribution - URL: https://www.eclass.eu/en/standard/eclassrdfowl - Retrieval tags: eclass, industrial-semantic-standards, ontology-governance, phase16, product-classification, rdf, semantic-interoperability, supply-chain The eCl@ss RDF/OWL representation exposes the eCl@ss product and service classification system as semantic-web resources. It supports product classification, properties, identifiers, and machine-readable product semantics across procurement and industry contexts. Key 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. ## 126. GraphDB documentation: reasoning, SHACL, and semantic graph database capabilities - Authors/Org: Ontotext - Year: 2026 - Venue/Site: Ontotext GraphDB Documentation - Bucket: technical - Quality: official_docs - URL: https://graphdb.ontotext.com/documentation - Retrieval tags: graphdb, implementation-evidence, ontotext, rdf, reasoning, semantic-retrieval, semantic-search, shacl, sparql Official documentation for GraphDB, an RDF database with SPARQL, inferencing, SHACL validation, repository management, and semantic retrieval features. Key claims: - RDF graph stores can apply inferencing to derive implicit relationships and classifications. - SHACL validation can enforce graph quality and business constraints. - Semantic retrieval can combine graph structure with text and similarity features. - RDF graph stores can apply inference to derive implicit classifications and relationships. - Semantic retrieval can combine graph structure, text, similarity, and provenance. ## 127. LlamaIndex Knowledge Graph RAG documentation - Authors/Org: LlamaIndex - Year: 2026 - Venue/Site: LlamaIndex Documentation - Bucket: technical - Quality: official_docs - URL: https://docs.llamaindex.ai - Retrieval tags: developer-tooling, knowledge-graph-index, llamaindex, property-graph, rag-framework Documentation for building knowledge graph and property graph indexes that can be used with retrieval-augmented generation workflows in LlamaIndex. Key claims: - Application frameworks expose KG indexes as retrievers for LLM context. - Graph extraction and graph query components can be embedded into common RAG stacks. - Developer ergonomics matter for adoption of ontology-like retrieval patterns. ## 128. Neo4j GraphRAG documentation and developer guidance - Authors/Org: Neo4j - Year: 2026 - Venue/Site: Neo4j Docs - Bucket: technical - Quality: official_docs - URL: https://neo4j.com/docs/neo4j-graphrag-python/current - Retrieval tags: cypher, graph-traversal, graphrag, llm, neo4j, property-graph, vector-search Neo4j documentation for building retrieval-augmented generation applications using graph data, vector indexes, Cypher retrieval, and graph-aware context expansion. Key claims: - GraphRAG combines vector retrieval with graph traversal or Cypher retrieval. - Graph neighborhoods can supply connected context that document chunks alone miss. - LLM answers can be grounded in graph paths and source nodes. - GraphRAG can combine vector similarity with explicit graph traversal or Cypher queries. - Graph neighborhoods provide connected context that chunk retrieval alone can miss. ## 129. Palantir Reports Q1 2026 U.S. Revenue Growth of 104% Y/Y and Revenue Growth of 85% Y/Y - Authors/Org: Palantir Technologies Inc. - Year: 2026 - Venue/Site: Palantir Investor Relations - Bucket: palantir - Quality: primary_source - URL: https://investors.palantir.com/news-details/2026/Palantir-Reports-Q1-2026-U-S--Revenue-Growth-of-104-YY-and-Revenue-Growth-of-85-YY-Raises-FY-2026-Revenue-Guidance-to-71-YY-Growth-and-U-S--Comm-Revenue-Guidance-to-120-YY-Crushing-Consensus-Expectations - Retrieval tags: aip, investor-relations, palantir, q1-2026, revenue-growth Q1 2026 earnings release reporting continued revenue growth and updated guidance, useful for commercial context around AIP adoption claims. Key claims: - Palantir reported Q1 2026 U.S. revenue growth of 104% year over year. - The release raises FY2026 revenue guidance and U.S. commercial guidance. - The release frames growth around ongoing demand for Palantir software and AI. ## 130. Palantir Reports Q4 2025 U.S. Commercial Revenue Growth of 137% Y/Y and Revenue Growth of 70% Y/Y - Authors/Org: Palantir Technologies Inc. - Year: 2026 - Venue/Site: Palantir Investor Relations - Bucket: palantir - Quality: primary_source - URL: https://investors.palantir.com/news-details/2026/Palantir-Reports-Q4-2025-U-S--Comm-Revenue-Growth-of-137-YY-and-Revenue-Growth-of-70-YY-Issues-FY-2026-Revenue-Guidance-of-61-YY-and-U-S--Comm-Revenue-Guidance-of-115-YY-Crushing-Consensus-Expectations - Retrieval tags: aip, commercial-growth, investor-relations, ontology-mcp, palantir, q4-2025 Q4 2025 earnings release and business update presenting Palantir's revenue growth, commercial growth, and AIP product/update narrative. Key claims: - Q4 2025 U.S. commercial revenue grew 137% year over year according to Palantir. - Palantir issued FY2026 revenue guidance and highlighted AIP product momentum. - Investor presentation materials mention Agents and Ontology MCP as new AIP tools. - Palantir reports rapid U.S. commercial revenue growth and positions AIP as a driver of demand. - The release is useful for adoption narrative but is not independent technical validation. ## 131. Palantir Technologies Inc. 2024 Form 10-K - Authors/Org: Palantir Technologies Inc.; U.S. Securities and Exchange Commission - Year: 2025 - Venue/Site: SEC EDGAR - Bucket: palantir - Quality: primary_source - DOI/Identifier: sec form 10-k, fiscal year ended 2024-12-31 - URL: https://www.sec.gov/Archives/edgar/data/1321655/000132165525000022/pltr-20241231.htm - Retrieval tags: 10-k, aip, enterprise-ai, foundry, gotham, palantir, sec Annual report describing Palantir's principal platforms, including Gotham, Foundry, Apollo, and AIP, and stating that AIP leverages generative AI and LLMs directly within Gotham and Foundry to operationalize AI on enterprise data. Key claims: - Palantir identifies Gotham, Foundry, Apollo, and AIP as principal platforms. - Foundry and Gotham transform information into integrated assets reflecting operations. - AIP uses machine learning and generative AI directly within Gotham and/or Foundry to operationalize AI on enterprise data. ## 132. PoolParty Semantic Suite product and documentation materials - Authors/Org: Semantic Web Company - Year: 2026 - Venue/Site: PoolParty / Semantic Web Company - Bucket: commercial - Quality: official_docs - URL: https://www.poolparty.biz/products/semantic-suite - Retrieval tags: entity-extraction, poolparty, semantic-ai, semantic-search, skos, taxonomy PoolParty materials describe taxonomy management, knowledge graph construction, semantic search, entity extraction, and semantic AI workflows for enterprise content and data. Key claims: - Controlled vocabularies and taxonomies can provide a pragmatic entry point to enterprise knowledge graphs. - Entity extraction and tagging link unstructured content to curated concepts. - Semantic expansion improves search, recommendation, and AI context quality. ## 133. RDF Mapping Language (RML) - Authors/Org: RML Community Group - Year: 2024 - Venue/Site: RML.io Specification - Bucket: technical - Quality: official_docs - URL: https://rml.io/specs/rml - Retrieval tags: data-integration, heterogeneous-data, knowledge-graph-construction, mapping-language, phase15, rdf, rml, semantic-interoperability The RML specification defines RML as a superset of R2RML for customized mapping rules from heterogeneous data structures and serializations to RDF. The specification documents logical sources, reference formulations, term maps, and source-agnostic mapping constructs. Key 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. ## 134. Semantic Interoperability in Data Spaces - Authors/Org: International Data Spaces Association - Year: 2024 - Venue/Site: International Data Spaces Association / Zenodo - Bucket: technical - Quality: official_docs - DOI/Identifier: 10.5281/zenodo.10964377 - URL: https://zenodo.org/records/10964377 - Retrieval tags: data-governance, data-sovereignty, data-space, dataspaces, idsa, phase15, semantic-interoperability, usage-policy The IDSA position paper explains why semantic interoperability is required in data spaces and frames common information models, metadata, vocabularies, and governance as necessary for trusted data exchange across participants. Key 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. ## 135. Solid Protocol - Authors/Org: W3C Solid Community Group - Year: 2025 - Venue/Site: Solid Project / W3C Community Group - Bucket: technical - Quality: official_docs - DOI/Identifier: solid protocol editor's draft - URL: https://solidproject.org/TR/protocol - Retrieval tags: access-control, decentralized-data, federated-data-sharing, linked-data, ontology-governance, phase16, solid, solid-pods The Solid Protocol specifies decentralized data storage through identity, authentication, authorization, linked-data resources, and storage servers called pods. It gives a standards-adjacent model for user-controlled linked data and app interoperability. Key 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. ## 136. SPARQL 1.2 Query Language - Authors/Org: World Wide Web Consortium - Year: 2026 - Venue/Site: W3C Working Draft - Bucket: technical - Quality: official_docs - DOI/Identifier: w3c wd-sparql12-query-20260617 - URL: https://www.w3.org/TR/sparql12-query - Retrieval tags: agent-tools, query-language, rdf, sparql-1-2, w3c Current W3C Working Draft for SPARQL 1.2 Query Language, the standard query language for RDF graph pattern matching and graph retrieval. Key claims: - Graph query languages are the operational access layer for semantic knowledge bases. - Standard query semantics enable tool interoperability and verifiable retrieval. - SPARQL 1.2 indicates ongoing evolution of RDF query capabilities. ## 137. Stardog Virtual Graphs documentation - Authors/Org: Stardog - Year: 2026 - Venue/Site: Stardog Docs - Bucket: commercial - Quality: official_docs - URL: https://docs.stardog.com/virtual-graphs - Retrieval tags: data-virtualization, enterprise-kg, implementation-evidence, rdf, semantic-layer, sparql, stardog, virtual-graph, virtual-graphs Documentation for mapping external data sources into Stardog as virtual RDF graphs, enabling SPARQL access and semantic integration without fully copying source data. Key claims: - Virtual graphs expose external enterprise data through semantic mappings. - Ontology-backed graph access can decouple business queries from physical schemas. - Virtualization can be combined with materialized graph data when needed. ## 138. The Role of Ontologies in Gaia-X - Authors/Org: Gaia-X European Association for Data and Cloud AISBL - Year: 2023 - Venue/Site: Gaia-X - Bucket: technical - Quality: official_docs - URL: https://gaia-x.eu/the-role-of-ontologies-in-gaia-x - Retrieval tags: compliance, data-space, gaia-x, json-ld, ontology-governance, phase15, policy-reasoning, shacl, verifiable-credentials Gaia-X documents the role of ontologies in compliance and policy reasoning. Gaia-X uses RDF triples, verifiable credentials, SHACL shapes, JSON-LD schemas, and OWL ontology artifacts to check consistency and support machine-readable trust and compliance operations. Key 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. ## 139. TopBraid EDG enterprise knowledge graph and data governance materials - Authors/Org: TopQuadrant - Year: 2026 - Venue/Site: TopQuadrant - Bucket: commercial - Quality: official_docs - URL: https://www.topquadrant.com/products/topbraid-edg - Retrieval tags: data-governance, ontology-management, shacl, taxonomy, topbraid-edg, topquadrant Product and documentation material for TopBraid EDG, covering enterprise ontology, taxonomy, reference data, data catalog, governance workflows, and SHACL-based validation. Key claims: - Ontology and taxonomy assets need governance workflows, stewardship, versioning, and publishing. - Business glossaries, reference data, taxonomies, and catalogs can be managed as connected semantic assets. - SHACL constraints can operationalize semantic governance. ## 140. YARRRML: Human Readable Text-Based Representation for Declarative Generation Rules - Authors/Org: RML Community Group - Year: 2025 - Venue/Site: YARRRML Specification - Bucket: technical - Quality: official_docs - URL: https://w3id.org/yarrrml/spec - Retrieval tags: data-integration, knowledge-graph-construction, mapping-language, phase15, rdf, rml, semantic-interoperability, yaml, yarrrml YARRRML is a human-readable YAML-based representation for declarative generation rules that can be converted to RML. It lowers authoring complexity for users defining RDF generation rules over heterogeneous sources. Key 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. ## 141. A Survey of Graph Retrieval-Augmented Generation for Customized Large Language Models - Authors/Org: Yuntong Hu; et al. - Year: 2024 - Venue/Site: arXiv - Bucket: academic - Quality: peer_reviewed_survey - DOI/Identifier: arxiv:2501.13958 - URL: https://arxiv.org/abs/2501.13958 - Retrieval tags: customized-llm, domain-adaptation, evaluation, graph-memory, graphrag, knowledge-graph, rag, retrieval, survey Survey of GraphRAG methods for customized LLMs, covering graph construction, graph retrieval, augmentation strategies, applications, and evaluation. Key claims: - GraphRAG methods differ in how they construct graphs, retrieve subgraphs, augment prompts, and evaluate answers. - Customized LLM systems need domain-specific graph construction and maintenance strategies. - Ontology can supply schema, constraints, and provenance that many graph-RAG pipelines otherwise generate weakly. - GraphRAG customizes LLMs by externalizing domain structure into graph memory. - Graph construction, retrieval strategy, and update mechanisms are central design variables. ## 142. Agentic Retrieval-Augmented Generation: A Survey on Agentic RAG - Authors/Org: Survey authors - Year: 2025 - Venue/Site: arXiv - Bucket: academic - Quality: peer_reviewed_survey - DOI/Identifier: arxiv:2501.09136 - URL: https://arxiv.org/abs/2501.09136 - Retrieval tags: agent-memory, agentic-rag, agents, evaluation, knowledge-graph, rag, tool-use Survey of agentic RAG systems that use agents for planning, query decomposition, tool use, memory, retrieval orchestration, and answer synthesis. Key claims: - Agentic RAG extends RAG with planning, tool use, memory, and dynamic retrieval strategies. - Evaluation should cover both final answers and intermediate retrieval/tool decisions. - Ontology and knowledge graphs can define retrievable objects, permissible actions, and provenance for agentic RAG. ## 143. Croissant: A Metadata Format for ML-Ready Datasets - Authors/Org: Mubashara Akhtar; Omar Benjelloun; Costanza Conforti; Luca Foschini; Pieter Gijsbers; Joan Giner-Miguelez; Sujata Goswami; Nitisha Jain; Michalis Karamousadakis; Michael Kuchnik; et al. - Year: 2024 - Venue/Site: ACM DEEM - Bucket: academic - Quality: peer_reviewed_survey - DOI/Identifier: 10.1145/3650203.3663326 - URL: https://doi.org/10.1145/3650203.3663326 - Retrieval tags: croissant, dataset-metadata, json-ld, machine-learning-data, ml-ready-data, phase14, responsible-ai, schema-org, semantic-interoperability Croissant is a metadata format for ML-ready datasets developed by an MLCommons community and published in the 2024 ACM DEEM workshop. It builds on schema.org/JSON-LD to describe dataset metadata, resources, structure, and ML semantics so datasets can be discovered, loaded, and reused across tools and repositories. Key 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. ## 144. Enterprise Digital Twins in Financial Services: Convergence of Architecture, Operations, and Engineering - Authors/Org: Glen Stokes - Year: 2025 - Venue/Site: paper - Bucket: palantir - Quality: scholarly_index - DOI/Identifier: 10.36227/techrxiv.175624549.98427022/v1 - URL: https://openalex.org/W4413638561 - Retrieval tags: openalex, palantir The Enterprise Digital Twin (EDT) represents the convergence of decades-long parallel innovations in enterprise architecture (Zachman, TOGAF, ArchiMate), systems engineering (Agile, DevOps), and operational management (ITIL, SRE, observability). This paper demonstrates that EDTs are not an abrupt invention but the culmination of six decades of theory and practice-from 1960s socio-technical systems to contemporary AI-enabled, graph-driven models. Recent advances in graph databases, semantic modeling, and AI-augmented engineering now make implementation feasible, with EDTs providing the context that is so important for agentic AI. The paper examines EDT foundations in ontology, taxonomy, and graph theory; reviews enabling standards and technologies; outlines socio-technical implementation methodology; and analyzes real-world applications including ING's capability transformation, Lloyds' observability model, and Singapore's national digital twin. FSIs can adopt EDTs with confidence, applying proven frameworks through modern technologies. The purpose of an EDT is to help FSIs to address one of their key challenges-accelerating innovation without compromising compliance and security-but the principles apply equally to healthcare and government. Starting to build an EDT presents an opportunity for architecture, engineering and operations teams to pool their strengths around a common business problem. We discuss the rise of an Autonomous Agentic Twin: orchestrated agentic swarms that cooperate with the EDT to autonomously execute business processes, make governance-compliant decisions, and adapt to changing conditions. This evolution represents the rapidly approaching future of autonomous enterprises. ## 145. ForestGPT and Beyond: A Trustworthy Domain-Specific Large Language Model Paving the Way to Forestry 5.0 - Authors/Org: Florian Sommer, Benno Eberhard, Andreas Holzinger - Year: 2025 - Venue/Site: Electronics - Bucket: palantir - Quality: scholarly_index - DOI/Identifier: 10.3390/electronics14183583 - URL: https://openalex.org/W4414098962 - Retrieval tags: openalex, palantir Large language models (LLMs) such as Chat Generative Pre-Trained Transformer (ChatGPT) are increasingly used across domains, yet their generic training data and propensity for hallucination limit reliability in safety-critical fields like forestry. This paper outlines the conception and prototype of ForestGPT, a domain-specialised assistant designed to support forest professionals while preserving expert oversight. It addresses two looming risks: unverified adoption of generic outputs and professional mistrust of opaque algorithms. We propose a four-level development path: (1) pre-training a transformer on curated forestry literature to create a baseline conversational tool; (2) augmenting it with Retrieval-Augmented Generation to ground answers in local and time-sensitive documents; (3) coupling growth simulators for scenario modeling; and (4) integrating continuous streams from sensors, drones and machinery for real-time decision support. A Level-1 prototype, deployed at Futa Expo 2025 via a mobile app, successfully guided multilingual visitors and demonstrated the feasibility of lightweight fine-tuning on open-weight checkpoints. We analyse technical challenges, multimodal grounding, continual learning, safety certification, and social barriers including data sovereignty, bias and change management. Results indicate that trustworthy, explainable, and accessible LLMs can accelerate the transition to Forestry 5.0, provided that human-in-the-loop guardrails remain central. Future work will extend ForestGPT with full RAG pipelines, simulator coupling and autonomous data ingestion. Whilst exemplified in forestry, a complex, safety-critical, and ecologically vital domain, the proposed architecture and development path are broadly transferable to other sectors that demand trustworthy, domain-specific language models under expert oversight. ## 146. Neo4j customer stories for enterprise graph and knowledge graph applications - Authors/Org: Neo4j - Year: 2026 - Venue/Site: Neo4j Customer Stories - Bucket: commercial - Quality: primary_source - URL: https://neo4j.com/customers - Retrieval tags: customer-stories, graph-analytics, knowledge-graph, neo4j, property-graph Neo4j customer stories cover graph applications in recommendations, fraud, master data, life sciences, supply chain, identity, and knowledge discovery. Key claims: - Property graphs are useful where business value depends on relationship traversal and pattern discovery. - Connected data models can power operational applications and analytics. - Customer stories demonstrate adoption breadth but often need independent validation for quantitative claims. ## 147. NHS Federated Data Platform privacy notice - Authors/Org: NHS England - Year: 2024 - Venue/Site: NHS England - Bucket: technical - Quality: primary_source - URL: https://www.england.nhs.uk/digitaltechnology/nhs-federated-data-platform/security-privacy/nhs-fdp-privacy-notice - Retrieval tags: data-governance, fdp, governance, health-data, nhs, palantir, personal-data, privacy-notice Official privacy notice describing processing of personal data in the NHS Federated Data Platform, including purposes, data sources, access, protection, retention, and legal grounds. Key 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. - The notice is central for assessing public claims about data use limits and legal basis. ## 148. Ontology-Based Data Access: A Survey - Authors/Org: Guohui Xiao; Diego Calvanese; Roman Kontchakov; Domenico Lembo; Antonella Poggi; Riccardo Rosati; Michael Zakharyaschev - Year: 2018 - Venue/Site: IJCAI - Bucket: academic - Quality: peer_reviewed_survey - DOI/Identifier: 10.24963/ijcai.2018/777 - URL: https://www.ijcai.org/proceedings/2018/0777.pdf - Retrieval tags: data-integration, obda, ontology-based-data-access, phase14, query-rewriting, semantic-layer, virtual-knowledge-graph The IJCAI survey presents ontology-based data access (OBDA) as a semantic paradigm for user-friendly access to data repositories, especially relational data. It covers the ingredients of OBDA, theoretical results, techniques, applications, and future challenges. Key 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. ## 149. RML: A Generic Language for Integrated RDF Mappings of Heterogeneous Data - Authors/Org: Anastasia Dimou; Miel Vander Sande; Pieter Colpaert; Ruben Verborgh; Erik Mannens; Rik Van de Walle - Year: 2014 - Venue/Site: LDOW 2014 - Bucket: academic - Quality: peer_reviewed_survey - URL: https://ceur-ws.org/Vol-1184/ldow2014_paper_01.pdf - Retrieval tags: data-integration, heterogeneous-data, knowledge-graph-construction, mapping-language, phase15, rdf, rml, semantic-interoperability RML extends R2RML from relational databases to heterogeneous sources and serializations. The LDOW paper introduces RML as a generic mapping language for integrated RDF mappings over CSV, XML, JSON, and other data sources. Key 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. ## 150. Trustworthy GraphRAG: A Survey - Authors/Org: Survey authors - Year: 2024 - Venue/Site: arXiv - Bucket: academic - Quality: peer_reviewed_survey - DOI/Identifier: arxiv:2501.02157 - URL: https://arxiv.org/abs/2501.02157 - Retrieval tags: evaluation, graphrag, knowledge-graph, privacy, provenance, robustness, trustworthy-ai Survey-style preprint synthesizing trustworthiness dimensions for GraphRAG, including reliability, explainability, privacy, robustness, fairness, and evaluation challenges. Key claims: - GraphRAG trustworthiness depends on graph construction quality, retrieval reliability, explanation, privacy, robustness, and evaluation. - Graph structure can improve traceability but also introduces new failure modes from extraction, entity resolution, and stale edges. - Ontology-grounded GraphRAG should be evaluated as a graph lifecycle plus generation system.