支持卫生健康数据要素定义及规则推理的知识图谱

本数据集来源于医学知识抽取与图谱构建科研项目，旨在整合临床医学、药学及检验学等多领域知识资源，构建系统化、可计算的医学知识图谱。数据主要基于公开医学文献、权威数据库及结构化医疗资料，通过大语言模型驱动的命名实体识别与关系抽取方法生成。其内容涵盖疾病、症状、病因、治疗、药物、并发症等核心医学概念及其关联关系，形成完整的实体与关系网络。本数据集文件均为CSV格式，由3个文件组成。其中data.csv文件为知识图谱三元组信息，采用整数编号形式对实体与关系进行统一编码。entity_dict.csv和relation_dict.csv分别为实体与关系的数据词典，提供实数据索引id与数据内容的对照。数据内容以三元组结构存储，用于表示知识图谱中实体与实体之间的关联关系，数据总量包含216,775,940条三元组记录,其中包含26,325例推理规则，文件合计大小为15.31G。This dataset originates from a research project focused on medical knowledge extraction and knowledge graph construction. It aims to integrate knowledge resources across multiple fields including clinical medicine, pharmacy, and laboratory medicine, and build a systematic, computable medical knowledge graph. The data is primarily sourced from publicly available medical literature, authoritative databases, and structured medical materials, and is generated using Large Language Model (LLM)-driven named entity recognition and relation extraction methods. Its content covers core medical concepts such as diseases, symptoms, etiologies, treatments, medications, and complications, along with their associated relationships, forming a complete entity-relationship network. All files of this dataset are in CSV format, comprising three files in total. The data.csv file stores the triple information of the knowledge graph, and uniformly encodes entities and relationships via integer numbering. The entity_dict.csv and relation_dict.csv are the respective data dictionaries for entities and relationships, providing a mapping between their index IDs and corresponding content. The data is stored in a triple structure to represent the associative relationships between entities in the knowledge graph. The total number of triple records is 216,775,940, including 26,325 inference rules, and the combined total size of all files is 15.31 gigabytes.

搜集汇总

背景与挑战

背景概述

该数据集是一个支持卫生健康数据要素定义及规则推理的知识图谱，来源于医学知识抽取项目，整合了临床医学、药学和检验学等多领域知识。它包含疾病、症状、病因等核心医学概念及其关联关系，以CSV格式存储，由3个文件组成，总计216,775,940条三元组记录和26,325例推理规则，文件大小为15.31G。