Research On Intelligent Question And Answering System About TCM Knowledge

The question and answering system adopts natural language as an interactive way to give concise and accurate answers to the questions raised by users.The use of the question and answering system for information retrieval meets the needs of modern society for rapid access to information,and the development of knowledge graph provides abundant and accurate data support for the realization of the goal of the question and answering system.The purpose of the knowledge based question answering system is to use the knowledge graph storing structured knowledge as the data support platform,based on this platform,to realize the data interaction between the questions in natural language form raised by the users and the knowledge graph,and return the answers.As a traditional Chinese medical science,TCM has accumulated profound and effective medical practice theory.The research of intelligent question and answering system about TCM knowledge mainly includes the following aspects:First of all,focusing on the disease,the crawler technology was used to extract effective knowledge information on the Chinese Medicine website,then entity recognition,relationship extraction and attribute definition were performed on the information in order to build a TCM knowledge graph including 10 entity types,9 relationship types and 1 attribute type.The second task was to analyse the problem raised by users.The main task was to extract the keywords of TCM and match the relationship or attribute of the questions inputted by users.According to the expression characteristics of the TCM term,the AhoCorasick automata algorithm was proposed to extract the keywords of TCM from the questions in natural language form.The algorithm has a positive effect on TCM keywords extraction.In the relationship or attribute matching module,the matching process of the relationship or attribute was studied basing on the template matching method.This method was simple and efficient to match the problem and the relationship or attribute in the TCM knowledge graph from the character level.Using Bi LSTM with siamese structure to model the semantic vector of the problem and TCM knowledge graph unit which could match the relationship or attribute from the semantic level.It effectively overcame the differences between the expression of questions and knowledge information in knowledge graph.As a result,the combination of the two methods was proposed to effectively implement the matching operation of relationship or attribute.Finally,after obtaining the TCM keywords and relationship or attribute in the questions,the question was converted into query language on the knowledge graph and information retrieval was carried out in the knowledge graph about TCM to obtain the answer and returned it to the users.The experimental results show that the intelligent question and answering system about TCM knowledge completely makes the process from data collection to question and answering retrieval application come true,and also answer the natural language questions of users accurately.Figure 33;Table 14;Reference 51...