Question And Answer System Of Shanxi Tourism Diet Based On Knowledge Graph

With the rapid development of the Internet,"informatization" is embodied in all aspects of People's Daily life.In the era of big data,people's life has been inseparable from the support of the Internet,and the development model of "Internet +" has become a trend."Internet + tourism" has also become the most common mode of operation and development in China's major scenic spots.As the number of tourists increases year by year,under the traditional tourism mode,it is increasingly difficult for tourists to obtain information of scenic spots in a timely manner,and a new way is urgently needed to meet the needs of tourists.In this context,combining with the application and development of natural language processing,a kind of intelligent question-answering system for users emerges.Compared with the traditional search engine,intelligent question answering system is more concise and direct.Users can quickly get the answers they need based on their questions.To some extent,this greatly improves the efficiency of retrieval,and users can better understand the information of the scenic spot,thus bringing users a better travel experience.This paper aims at the field of tourism,taking Shanxi Province as an example,an intelligent question answering system based on BILSTM-CNN-CRF model in deep learning is presented.The main work of this paper is as follows:(1)With the Scrapy crawler framework in Python,I crawled the Shanxi tourism data from Ctrip,Tuniu and Qunar,integrated all the data and imported it into Neo4 j to construct the stealth knowledge graph of Shanxi tourism.(2)On the basis of most BILSTM-CRF question answering models based on knowledge graph,a layer of CNN is added to propose the question answering model of BILSTM-CNN-CRF,which makes the knowledge graph constructed in this paper more accurate and efficient in entity relationship recognition.Then,the Attention mechanism and CNN are respectively cited to obtain the connection between questions and relations.Finally,the Droppout strategy is introduced to prevent data set fitting.Finally,the data set was evaluated,and the predicted entity relationship pair and relationship recognition module were analyzed experimentally respectively.The results showed that the model method in this paper was significantly better than the traditional single-structure model and the BILSTM-CRF model.(3)On the basis of the above,an intelligent question-and-answer system for Shanxi tourism diet based on BILSTM-CNN-CRF model was constructed.