User Interest Mining Based On E-commerce Data Then Recommendation POI According User Interest

With the rapid development of information technology and the increase of access to information,people can get a large amount of information and data anytime and anywhere.The problem then is to choose the information that suits one's preference from the huge amount of information data.Recommendation system is a commonly used means to solve information overload.It can play a role in various fields with other technologies to improve the quality and efficiency of users' access to information,and bring users a better experience.At present,in various map platforms,such as Baidu Map and Gaode Map,they mainly provide common location query services,but they do not provide personalized POI recommendation services for users according to their personalized needs.In this paper,LDA topic model will be used as a bridge between users and POI to realize personalized POI recommendation for users according to their interests.The main work is as follows:(1)For solving the problem that some users produce less operation behavior in user behavior data,which is insufficient to analyze the user interest topic,cooperative filtering algorithm has been used to mine the potential products that users may be interested in by using user behavior data.(2)For the problem of lack of commodity information in user behavior data,detailed commodity information has been obtained by using web crawler technology to compensate for the lack of commodity information in corpus.(3)Integrating user and commodity information obtained before,and using web crawler to obtain POI information,a corpus has been constructed as the basic data in the experimental research.(4)The LDA topic model has been constructed and used to analyze the corpus to obtain users' preferred topics and POI topics.They have been used to evaluate the accuracy of the built topic model.Then a personalized POI recommendation algorithm based on LDA topic model has been proposed.(5)The design and development of POI personalized recommendation prototype system based on user data of e-commerce platform are completed.User behavior data mining based on e-commerce platform is realized.LDA topic model is used to analyze the topic of user interest and POI,and JS(Jensen-Shannon)distance is used as an index to measure the similarity between user interest and POI,so as to provide users with personalized POI recommendation function in map platform.It is shown that it is feasible to use the LDA topic model as a bridge between users and POI to realize the personalized POI recommendation function for users.And this method can avoid the problem that the user-poi scoring matrix is too large and sparse for the algorithm to carry out the POI recommendation research based on the traditional collaborative filtering algorithm and the deep learning algorithm.LDA subject model is used as a bridge to realize the cross-platform use of e-commerce data on the map platform.It has a certain engineering reference value for realizing personalized recommendation services for users according to personalized needs,and has potential reference value for exploring the law of human behavior data mining using cyberspace and providing intelligent location-based services.