Design And Implementation Of User Profile System Based On Text Mining

With the continuous development of Internet technology,the information generated by the network is constantly increasing.The accumulation of massive data and information leads to the inundation of effective information,which not only makes it difficult for users to reach effective information but also leads to the decreasing transformation of valuable users in the enterprise platform which decline the profits of enterprise.Therefore,it is very important to build a user profile system that can help the enterprises to understand users,locate users and tap users' features.The main goal of the paper is to design and implement a user profile system that can dig into user's interest feature and help the enterprise to locate users.In view of this goal,the main work is as follows.This paper investigated the current user profile construction methods and the technology of text mining.Based on the above work,a scheme of user profile system based on text mining has been designed and implemented.The data crawled from Douban city website is used as the data source of the profile system.Through data cleaning,word segmentation and feature selection,the data with distinct characteristic was obtained.Then,the topic interest of the platform was extracted by LDA topic model,and a comprehensive and abundant hierarchy user label system was built by combining with the method of domain ontology.In consideration of multiple characteristic factors,K-Nearest Neighbor Algorithm was used to map users with reasonable labels.K-means algorithm was used to find the user groups with similar features.In order to promote the social function of the platform,the expansion function of searching users with similar interests has been designed and realized.Based on Spring MVC and MyBatis backend framework which can convenient and decouple the business development process is convenient and decoupled,and the Bootstrap and Echarts framework which can form an intuitive and concise user profile visualization effect,a user profile system was implemented for comprehensive and meticulous positioning user interest characteristics.Meawhile,to verify the effect of system matching labels for users,the accuracy verification experiments was designed and carried out.