Research On Semi-automatic Construction Of Chinese Medical Knowledge Graph

"Chinese Internet users demand search behavior reporting of science" was released by Baidu index.The report found that people are most concerned about "Health & Medical" science theme.The upper applications base on The Knowledge Graph(KG)technology,such as Semantic Search,Query Understanding and Automatic Question Answering,they will be good help users to quickly get interest information and meet user's knowledge required from the mass of unstructured text.Therefore,this paper studies on semi-automatic construction of Chinese medical knowledge graph,hoping to find an effective method to build Chinese medical knowledge graph,helping construct domain knowledge graph to solve the problem about the upper application rely on the underlying knowledge,providing effective,more completion and reliable medical information for the user from the level of knowledge.The main achievements are as follows:1.Because the medical websites are usually characterized of storing entity and attribute values in different pages,we propose an automatic extraction system which is related to entity and attribute relations(attributes and values)of separate storage.This paper presents the relations about the attributes which are stored in many pages by effective annotation,then generates rules for data records extraction.Experiments have shown that the system can not only complete attribute relations of separate storage extraction,also be compatible with regular relation extraction,while maintaining high accuracy.2.For the issue that description information of the attribute relations of the entity data includes unstructured,we propose a method that based on Convolutional Neural Networks(CNNs)for Weak-surpervision relation extraction to extract relations of entities,and implement knowledge completition.The main method is to convert description data of entity to the sentence-level entity description,generate a corresponding relationship between the entity and its attributes as sample training data based on weak supervision method,complete the relation classification based on convolutional neural network method,reach the goal of relation extraction.Experimental results show that the method has high accuracy.3.Complete the design and implementation of system architecture of automatic construction for Chinese medical knowledge graph,and provide the visual interface of medical information for query and display.