2026-10-11T03:14:13Z https://www.scidb.cn/oai

10.57760/sciencedb.j00133.00733 2026-09-20T09:57:28Z

2026-09-20 Research Data and Code for Knowledge Indexing and Structured Analysis Methods for Mitigating Hallucinations in Large Language Models doi:10.57760/sciencedb.j00133.00733 en This dataset and code repository support the research paper titled "Knowledge Indexing and Structured Analysis Methods for Mitigating Hallucinations in Large Language Models". It includes:Source Code & Algorithms: Implementation of the BERT-CRF domain concept indexing, head-concept-driven dynamic window relation extraction, and TextRank centrality calculation scripts.Experimental Prompts & Configs: Structured Prompt templates and Structured Retrieval-Augmented Generation (RAG) configuration files used in Section 4.2 for GPT-4o (temperature=0, max_tokens=512).Dataset Samples: Expert-annotated concept indexing and relation triplet samples in the diabetes domain (including schema descriptions for 1,458 training and 369 validation samples). Large Language Models; Knowledge Enhancement; Knowledge Indexing; Structured Analysis; Hallucination Mitigation liu ya jun PUBLIC https://creativecommons.org/licenses/by/4.0/ dataset Science Data Bank