LLMs autonomously evolve ontologies with dual-decoder framework

Researchers have developed Evo-DKD, a novel framework designed to autonomously evolve ontologies and knowledge graphs using Large Language Models (LLMs). This dual-decoder system combines structured ontology traversal with unstructured text reasoning, generating both ontology edits and natural-language justifications. Evo-DKD operates in a closed loop, validating proposed edits before integrating them into the knowledge base to inform future reasoning. Experiments demonstrate its effectiveness in refining healthcare ontologies, improving semantic search, and modeling cultural heritage timelines, outperforming existing methods. AI

IMPACT Offers a new paradigm for LLM-driven knowledge base maintenance, combining symbolic and neural reasoning for sustainable ontology evolution.

RANK_REASON The cluster contains an academic paper detailing a novel method for LLM-driven ontology evolution. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• Evo-DKD
• Large Language Models
• LLMs

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →