摘要
知识图谱作为知识的结构化表示,已成为智能认知系统的一个重要研究方向。本文主要通过抓取网络上以骨科康复治疗知识为主的各种医学知识,通过数据清洗将这些医学数据整合成以疾病为中心的结构化知识。同时辅助权威书籍和医生专家对数据进行修正和增删。最终将数据存入图数据库,构建七类约9.6万的实体节点、十类约109万的实体关系以及九类属性类型的医疗健康知识图谱。本文进一步对构建的知识图谱进行通用疾病和骨科疾病分析,发现实体关系边数分布满足幂律分布,同时分析了各类型实体关系边数排行榜,为病人食谱、症状、检查项目和并发症等选择提供一定的参考价值,本知识图谱也为骨科康复的互联网远程智能问诊打下基础。
Knowledge Graph as a structured representation of knowledge has become an important research topic for intelligent cognitive systems. In this paper, various medical knowledge dominated by knowledge of orthopedic rehabilitation treatment on the web is captured, and these medical data are integrated into structured disease-centric knowledge through data cleansing. At the same time, corrections, additions and deletions to the data are made through authoritative books and physi-cian experts. Finally, medical data is deposited into the Graph Database, we construct seven cate-gories of approximately 96,000 entity nodes, ten categories of approximately 1.09 million entity relationships, and nine categories of attribute types for medical health Knowledge Graph. We fur-ther analyze the constructed Knowledge Graph for generalized diseases and orthopedic diseases, and find that the distribution of entity relationship edges satisfies the idempotent distribution. We also analyze the ranking of entity relationship edges for each type of entity relationship, which provides some reference value for the selection of patient recipes, symptoms, examination items and complications, etc. This Knowledge Graph also lays a foundation for internet remote intelligent consultation for orthopedic rehabilitation.
出处
《人工智能与机器人研究》
2020年第3期182-193,共12页
Artificial Intelligence and Robotics Research
关键词
知识图谱
图数据库
骨科康复
Knowledge Graph
Graph Database
Orthopedic Rehabilitation
