News Recommendations Combining User Weibo Interest Mining And Collaborative Filtering

The highway as the information of the Internet has experienced rapid development,and the "information explosion" brought about by the development process has seriously affected the experience of users using the Internet.The recommendation system came into being under the requirements of efficient information retrieval.The recommendation system uses the existing data information to recommend information that may be of interest to the user through various algorithms,and then adjusts the recommendation direction through feedback of the user's recommendation information,and continuously optimizes the recommendation system.In practical applications,the "cold start" based on content recommendation in the recommendation system and the lack of potential interest arise.The introduction of existing information data outside the user in the recommendation system is one of the solutions to the problem of cold start and potential loss of interest.The disclosure,accuracy and information diversity of personal social media(microblogging)information are widely studied in personality.Recommendation system.At the same time,the collaborative filtering algorithm has been widely studied as a method to solve the lack of recommendation diversity.The basic idea of collaborative filtering algorithms is to recommend based on user behavior.In recent years,there has been more and more research on social media.Researchers analyze social media for hot news forecasting,public opinion analysis,personalized recommendation and community discovery.By tapping social media information,personal images of individualized groups are imaged by individuals,and individual user images,microblog posts,forwarding and comments are analyzed as the basis for personalized recommendations.In this paper,the interest mining of individual users is realized by studying the microblog media structure,and the mining user interest set is used as one of the metadata based on content news recommendation.According to the above research ideas to solve the problem of insufficient cold start and recommendation diversity in the actual news recommendation application process,The main content of this article is as follows:Firstly,aiming at the cold start problem of traditional recommendation algorithm,a news recommendation algorithm combining user microblog interest mining and collaborative filtering is proposed.This algorithm calculates the user's microblog data and mines user interest.Based on the candidate recommendation news set of user microblog interest mining,the user history news evaluation information is used,and the user-based collaborative filtering algorithm is used to obtain the final candidate news set,thereby solving the problem of cold start and recommendation diversity missing,and improving the recommendation effect.Secondly,in the process of mining user microblog interest,through the analysis of group users,Weibo found that there are many microblog users with less microblogs.In response to this problem,the microblog interest mining framework is used to mine the microblog user followers.The background information and label information are used to express the interest of microblogging users.A microblog user interest mining algorithm is proposed.The algorithm integrates user interest,user potential interest and key interest mining,and performs microblog portraits on users.Better build.Finally,the dimensional disaster problem occurred in the process of news recommendation.Text classification was carried out through textCNN,which greatly reduced the computational consumption time in the recommendation process.At the same time,compared with the traditional news recommendation algorithm,the experiment was compared and analyzed.The experimental results show that the proposed algorithm has the diversity and novelty while improving the recommendation effect,and can effectively alleviate the cold start problem of new users.