EGT-KG: Evidence-Grounded Typed KG Retrieval for Practical Scientific QA with Small Language Models

Researchers proposed the Evidence-Grounded Typed Knowledge Graph (EGT-KG), a retrieval framework that improves scientific question-answering with local Small Language Models (SLMs). In experiments on a Biopolymer-bound Soil Composite literature benchmark, EGT-KG outperformed vanilla Retrieval-Augmented Generation (RAG) in most settings, with the best improvement from llama3:8b achieving a Final Score of 70.37 (+14.67%) and 68.82 (+12.14%) using automatically generated and expert-defined relation schemas, respectively.

arXiv:2609.00479v1 Announce Type: new Abstract: For emerging scientific research domains, local Small Language Models SLMs are becoming more attractive, as they offer stronger privacy control and more stable deployment pipelines than Large Language Models. However, in practice, scientific question-answering on SLMs often operates under inevitable constraints: small literature collections, fragmented evidence, limited context window and reasoning abilities. We propose the Evidence-Grounded Typed Knowledge Graph EGT-KG , a retrieval framework to improve information retrieval with local SLMs. We assessed three question-answering settings: a vanilla Retrieval-Augmented Generation RAG workflow and two EGT-KG workflows: an automatically generated relation schema AS and an expert-defined relation schema ES . Our experiments were evaluated with a six-dimensional evaluation framework S3CRF: Soundness, Correctness, Completeness, Conciseness, Relevance, Fluency on a Biopolymer-bound Soil Composite literature benchmark, showing that EGT-KG outperforms the vanilla RAG method in most settings, with the best improvement from llama3:8b: a Final Score of 70.37 +14.67% and 68.82 +12.14% by AS/ES EGT-KG variants.