knowledge-graphs

Andrej Karpathy has joined Anthropic to lead a team using Claude to accelerate pretraining research. The article also references Anthropic's published Claude cookbook on knowledge graph construction.

The paper proposes Stochastic Student Knowledge Graphs (SSKG) to enable faithful simulation of student knowledge by LLMs, overcoming the limitation where LLMs tend to perform at their own capability level instead of simulating varying mastery profiles.

This paper proposes a composable trust infrastructure for manufacturing knowledge graphs, focusing on cross-system provenance, temporal reasoning, and decision traceability.

This paper proposes a categorical approach to generate falsifiable research ideas by modeling papers as small categories and using a functor-preservation gate to filter cross-domain analogies, improving over traditional LLM-based methods.

Anthropic hired an engineer specialized in building knowledge graphs for multi-agent systems, and a 15-minute workshop demonstrates how to construct one using the Claude Agent SDK with efficient steps.

G-MARK is a grounded multi-agent reasoning framework that uses knowledge graphs to enhance cooperative driving by preserving object provenance and reducing communication payload while improving reasoning and planning accuracy.

This paper introduces RDFdL, a framework integrating RDF with Differential Dynamic Logic to enable reasoning about static knowledge and dynamic behaviors in cyber-physical systems.

ReLTEx is a framework for reliable LLM-based taxonomy expansion that combines LLM-driven candidate generation with structure-aware validation and recursive expansion control to reduce hallucinations and improve consistency. Experiments on benchmark taxonomies show it produces more reliable and semantically coherent expansions.

TRACE is a proposed retrieval-augmented conversational engine for public service chatbots that improves constraint-aware recommendations by strengthening retrieval quality over noisy directories, reducing hallucinated responses.

Introduces NL2SHACL-Bench, a benchmark suite for translating natural language requirements into SHACL shapes, evaluating four state-of-the-art LLMs and showing they struggle with semantic equivalence for complex patterns.

A technical post explaining why RAG fails for multi-hop queries and presenting a 9-step roadmap for building context graphs, where entities are nodes and relationships are edges, with a minimal 150-line engine.

KnowPlan is an extraction-first framework for personalized academic degree pathway planning, using CatalogBrowse to reconstruct curricula from heterogeneous university sources and DegreeMap to optimize personalized plans via CP-SAT, achieving high recall and feasibility across evaluations.

This paper proposes the Mecellem semantic protocol, an ontologically grounded framework for artificial legal intelligence, arguing that legal reasoning requires dynamic, context-dependent meaning construction rather than mere codification or statistical pattern recognition.

ProPRL introduces a property-aware framework for prerequisite relation learning in educational knowledge graphs, combining concept-resource hypergraph and directed behavior graph with adaptive pair-conditioned fusion and an irreversibility constraint to achieve state-of-the-art performance.

PULSE is a new executable contract language for spatiotemporal knowledge graph engineering, providing a typed runtime with role-based write effects, safety properties verified in Lean 4, and trace parity across large datasets.

This paper presents a methodology for integrating knowledge graphs into ROS 2-based robotic systems to improve autonomy and decision-making in missions like search and rescue, demonstrated with the Aerostack2 framework.

IRIS is a training-free framework that uses frozen large language models to construct reusable identity representations for entities in knowledge graphs, enabling efficient entity alignment across different KGs without pair-dependent processing.

HyCE-RAG is a novel hypergraph-based retrieval-augmented generation framework for multi-hop question answering that constructs explicit evidence chains via confidence-aware heuristic search, outperforming standard RAG and graph-based RAG methods in accuracy, relevance, and faithfulness.

IBM has launched a free 1-hour course on building agentic knowledge graphs, covering topics from introduction to multi-agent orchestration.

MA-DAR is a plug-and-play framework that addresses representation conflicts in replay-based continual temporal knowledge graph reasoning by aligning replayed and current representations on a shared manifold and using a dynamic gating mechanism for adaptive fusion.