mlsanityData to 14 Oct 2026

The Compressive Knowledge Graph Hypothesis: Which Graph Facts Matter for Scientific Hypothesis Generation?

arXiv 2605.27176 ↗First posted 26 May 2026cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

Knowledge graphs (KGs) can provide structured scientific context to language models, but it remains unclear which graph facts actually shape the generated hypotheses. We study KG-guided hypothesis generation for battery materials across Mistral-7B, Llama-3.1-70B, and Gemini 2.5 Flash. We perturb local KGs by varying density, ontology richness, topology, and control structure, and evaluate outputs with both provided-graph and fixed-reference metrics. Across models, KG utility is selective and model-dependent: graph context changes outputs, but no-KG outputs also recover substantial graph content from model priors. Compact top-k subgraphs often approximate full-KG behavior, including when claimed-outcome triples are held out. At the same time, compression is not unique to one semantic ranking rule, random and topology-based subsets can also recover much of the signal. These results support a redundancy-aware Compressive KG hypothesis: useful KG signal is often recoverable from compact, scientifically structured subgraphs rather than requiring the full local graph.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 19 of its 37 references that are in the map. First posted on arXiv 26 May 2026.

Builds on

19 of 37 references.

References · 19

  1. 2025Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilitiescited by 3,796
  2. 2023Lost in the Middle: How Language Models Use Long Contextscited by 3,455
  3. 2023Unifying Large Language Models and Knowledge Graphs: A Roadmapcited by 668
  4. 2024Graph Retrieval-Augmented Generation: A Surveycited by 324
  5. 2019K-BERT: Enabling Language Representation with Knowledge Graphcited by 304
  6. 2019Knowledge Enhanced Contextual Word Representationscited by 296
  7. 2024Retrieval-Augmented Generation with Graphs (GraphRAG)cited by 172
  8. 2022GreaseLM: Graph REASoning Enhanced Language Models for Question Answeringcited by 171
  9. 2022Deep Bidirectional Language-Knowledge Graph Pretrainingcited by 156
  10. 2024GNN-RAG: Graph Neural Retrieval for Large Language Model Reasoningcited by 128
  11. 2023Chain-of-Knowledge: Grounding Large Language Models via Dynamic Knowledge Adapting over Heterogeneous Sourcescited by 122
  12. 2024Graph Chain-of-Thought: Augmenting Large Language Models by Reasoning on Graphscited by 119
7 more
  1. 2025Retrieval-Augmented Generation with Conflicting Evidencecited by 63
  2. 2025Hypothesis Generation for Materials Discovery and Design Using Goal-Driven and Constraint-Guided LLM Agentscited by 30
  3. 2023Rationalization for Explainable NLP: A Surveycited by 22
  4. 2025Knowledge Graph-extended Retrieval Augmented Generation for Question Answeringcited by 14
  5. 2025EPERM: An Evidence Path Enhanced Reasoning Model for Knowledge Graph Question and Answeringcited by 7
  6. 2024Feature Attribution with Necessity and Sufficiency via Dual-stage Perturbation Test for Causal Explanationcited by 6
  7. 2025Recent Advances and Future Directions in Literature-Based Discoverycited by 1

Built on by

Not cited yet.