基于知识图谱的医疗智能对话机器人

Medical intelligent dialogue robot based on knowledge mapping

“天猫精灵”、“小度”等对话机器人能理解人类语言做出简短问答,但常常“答非所问”。为实现真正智能问答,提高对话的复杂度、准确性和可信度,设计和开发了基于知识图谱的医疗智能问答机器人。研究使用Apache-jena存储知识、RDF三元组表示知识、Jieba分词处理和Echarts构建知识图谱,基于Python Django框架和Apache-jena-fuseki图数据库开发疾病药理知识问答系统。系统接入161医院在线问诊小助手平台,经过实验测试,完成医患智能交互问答,提高诊断效率并方便病人。拓展行业数据集构建大型知识图谱和完善系统后,可在更多领域服务人类。

Dialogue robots such as“tmall genie”and“Xiaodu”can understand human language and do short question and answering(Q&A),but often“answer the wrong question”.In order to realize real intelligent Q&A and improve the complexity,accuracy and credibility of dialogue,a medical intelligent dialogue robot based on knowledge mapping is designed and developed.This paper studies the use of Apache Jean to store knowledge,RDF triplet to represent knowledge,Jieba word segmentation and Echards to construct knowledge mapping,and develops a disease pharmacology knowledge Q&A system based on Python Django framework and Apache-Jena-fuseki diagram database.The system is connected to the online consultation assistant platform of 161 hospital.After experimental test,it completes the doctor-patient intelligent interactive Q&A,improves the diagnosis efficiency and facilitates patients.After expanding the industry data set,building a large-scale knowledge mapping and improving the system,more fields can be applied to serve the human being.