Research On Shaanxi Tourism Question Answering System Based On Knowledge Graph

In today's information age,people's life can't do without various information.Now,it has become people's ordinary life state to obtain information on the internet.But with the continuous accumulation of quantity of information,the difficulty for users to select effective information also increases.The method of standardizing massive internet data and excavating its effective information to facilitate people's information retrieval is deemed as an important topic in information retrieval field.The occurrence of knowledge graph provides new knowledge carrier for standardized data.With the help of knowledge graph,the search results in information retrieval field such as search engine are greatly optimized in the form of display,the quality of display and the user experience degree.Knowledge graph as the question-answering system of knowledge library can also provide users with more accurate and convenient information retrieval service more intuitively.The field knowledge graph and the question-answering system with such knowledge graph as knowledge library are constructed based on the Shaanxi tourism field to meet the informational retrieval demands of tourists in tourism field.The main work of thesis is as follows:1.Website is provided for different data.Three kinds of data scraping schemes are designed and the Python's Scrapy crawler frame and Selenium's automation tools are applied to collect multi-source scenic spot data.In the meanwhile,the arrangement and secondary screening are conducted to the collected data.Then,Shaanxi tourism knowledge graph is constructed and stored in the Neo4J database by reference of the entity modeling method of Stanford University in medical domain.2.Tourism knowledge graph is used as knowledge library to construct the knowledge question and answer model;The fast text classification model Fast Text is applied in the construction of question and answer model,and a kind of Fast Text improvement scheme based on word feature fusion is proposed for the classification environment of short text in Chinese questions.Finally,experiment verifies that the classification effect of Chinese short text in the improved Fast Text is certainly improved,and the efficiency of question and answer model in handling with users' questions can also be improved.3.On the basis of knowledge graph and the question and answer model constructed in the above,the Shaanxi travel quiz assistant is developed,including the mobile terminal application development based on Flutter technology,the development of web server and the achievement of question-answering system.