🤖 AI Summary
Existing large language model–driven approaches to knowledge graph construction typically rely on stateless batch processing, which struggles to model cross-document semantic relationships, achieve precise entity disambiguation, and provide sufficient interpretability—limitations that hinder their deployment in high-stakes scenarios. To address these challenges, this work proposes the RAGA framework, which enables autonomous, end-to-end knowledge graph construction through a cognitive constraint mechanism of Read-Search-Verify-Construct and an atomic CRUD toolset. RAGA innovatively integrates ReAct-style agents, hybrid symbolic-vector retrieval, synchronized knowledge graph–vector updates, and evidence-anchored verification to substantially enhance the interpretability and traceability of the construction process. Evaluated on the QASPER scientific question answering subset, the method outperforms zero-shot baselines, significantly improving both answer accuracy and evidence quality.
📝 Abstract
Existing LLM-driven knowledge graph (KG) construction methods predominantly employ stateless batch processing pipelines, exhibiting structural deficiencies in cross-chunk semantic relation capture, entity disambiguation, and construction process interpretability. These limitations undermine KG quality, retrieval precision, and deployment trust in high-stakes domains.
We propose RAGA (Reading And Graph-building Agent), an LLM-based autonomous KG construction and retrieval fusion framework. RAGA provides an atomic toolset supporting full KG lifecycle CRUD operations and embeds a Read-Search-Verify-Construct cognitive constraint into a ReAct tool loop. A KG-vector synchronization mechanism enables hybrid symbolic-vector retrieval, while evidence-anchored verification links every knowledge entry to its source text for auditable provenance.
Preliminary experiments on a subset of the QASPER scientific QA dataset indicate that RAGA's fusion retrieval outperforms zero-shot baselines, with KG integration providing measurable gains in both answer and evidence quality. The framework design and experimental baseline serve as a reference for agent-driven autonomous KG construction.