knowledge-graph

The paper introduces an ontology-driven framework to quantify and enforce structural consistency in document-level relation extraction datasets, reducing logical contradictions and improving model generalization when using distant supervision data.

LingShu is a large-scale symptom-centric contextualized knowledge graph that bridges Traditional Chinese Medicine and modern biomedicine by integrating multi-source data to represent conditional medical knowledge.

This paper proposes the Defactualize-Steer-Rehydrate (DSR) framework, integrating knowledge graphs with activation steering to enhance style-controllable and fact-preserving generation in agentic conversational AI, evaluated on LLaMA models with significant improvements in factual recovery.

The paper introduces a knowledge-guided agentic framework that identifies missing patient context in health queries and asks targeted follow-up questions to improve the accuracy and consistency of downstream language model responses.

CTIFoundry introduces an agent-native corpus scaffold for cyber threat intelligence that improves LLM agent performance through structured ontology graphs and procedural skills, achieving higher accuracy and efficiency in investigations.

CoAL-RAG introduces a complexity-aware retrieval-augmented generation method for legal consultation that dynamically selects retrieval strategies based on question complexity, achieving significant improvements on Chinese and English legal benchmarks.

The paper proposes LUNG-KGMM, a knowledge-guided multimodal framework for predicting lung cancer incidence by integrating EHR, radiology data, and clinical guidelines, demonstrating superior performance on the MIMIC dataset and real-world validation.

GitNexus is an open-source knowledge graph tool for repositories that enhances coding agents by providing precise context, resulting in improved task success rates and cost efficiency.

This paper introduces MOOSEDev, a neurosymbolic memory system for coding agents that uses an ontology to structure project knowledge, demonstrating superior retrieval performance over vector-memory tools in handling architectural decisions and rationale.

MathCode is an AI-powered coding assistant that converts mathematical problems into Lean 4 theorems and attempts formal proofs, featuring a persistent REPL, theorem libraries, and agent-mode proving.

AstraZeneca describes Research Assistant, an internal LLM-based multi-agent system that lets scientists explore biomedical data via chat, with evidence grounding and citation links, deployed to 15,000 internal users.

This paper proposes SAG, a SQL-retrieval augmented generation architecture that organizes documents into event-entity hyperedges without building a global knowledge graph, enabling query-time dynamic linking of evidence chunks for multi-hop QA. It reports state-of-the-art retrieval and QA performance on HotpotQA, 2WikiMultiHopQA, and MuSiQue benchmarks.

This paper introduces BEST-KAG, a multimodal knowledge-driven framework for question answering on building engineering standards, leveraging a multimodal knowledge graph and graph-retrieval-based generation to improve clause-grounded traceability and outperform multiple LLMs.

Introduces the open-source project Semantica, which provides traceable decision-making basis and audit-support capabilities for AI Agents by building Context Graph / Knowledge Graph, suitable for scenarios with high explainability requirements such as finance and healthcare.

AVA-Encoder learns structured video representations via agentic auto-encoding with knowledge graphs, enabling cinematic video generation and reasoning while reducing token usage.

Introduces Mechanist, an autonomous agentic system that uses AI to discover and control the mechanisms underlying model intelligence, generating hypotheses, performing causal interventions, and improving safety and performance.

KGCache is an in-memory cache for one-hop knowledge graph neighborhoods that reduces redundant subgraph retrieval in KGQA systems with LLMs. Evaluated on WebQSP and CWQ, it achieves up to 1.91x faster KG retrieval and shows semantic caching further improves hit rates.

This paper introduces TKFQA, a counterfactual benchmark of 10,130 QA pairs over tables, texts, and knowledge graphs for evaluating LLM factuality consistency and order-robust reasoning, and proposes ORLF, a training framework that improves reasoning-chain accuracy and reduces input-order sensitivity.

This paper presents a self-improving 'researcher agent' for Text-to-SPARQL question answering over knowledge graphs, which iteratively refines its own prompts and tools. Evaluated on DBpedia, it achieves 0.22 accuracy and identifies predicate selection as the main bottleneck.

This paper proposes a hybrid knowledge graph generation pipeline that combines top-down grounding in Wikidata with bottom-up agentic synthesis to handle noisy, multilingual HR skill declarations, producing a scalable and self-healing skills taxonomy.