Research On Personalized Public Information Service Based On User Focusing And Recommendation Technology

The rapid development of the Internet provides an effective communication environment for the dissemination of public information,so that public information can be widely disseminated in a timely and timely manner by means of Internet carriers.Public information is the link that connects all aspects of society and plays an important role in the life and work of the public.In particular,the emergence of new media such as today's headlines,WeChat public account,Weibo,WeChat,etc.is closely integrated into the daily life of the public.The main medium and means of public information transmission.However,the public generates a large amount of data information in the process of using the Internet.In the process of using new media to spread information,information overload and dissemination information have low matching with user interest requirements.How to improve the efficiency of public information dissemination and make it highly efficient? The quality of the target audience? This article focuses on this issue and conducts the following series of related studies.This paper chooses to use the personalized information push mode of social network in the public information flow mode as the direction,and based on the typical microblog platform of social network new media,to explore the information acquisition mode of personalized push through social network(Microblogging).The main work is as follows:(1)Using Python web crawler to collect user data of Weibo platform,and performing preprocessing operations such as noise reduction,word segmentation and stop words on the collected data,and use this data as a corpus for subsequent model experiments.data.(2)Mining user interest based on LDA(Latent Dirichlet Allocation)topic model.LDA is a document topic generation model that can be used to identify potential topic information in large document sets or corpora.Based on the user data of Weibo platform,this paper analyzes the application status of LDA topic model in depth.Based on the LDA theme model and the characteristics of microblog text data sparseness,when constructing the three-layer Bayesian structure under the LDA model,the single user is merged.The historical blog set replaces the document layer,describes the user's interest with the probability distribution of the user under each theme,and mines the user interest theme.(3)The optimized K-means clustering method is used to cluster the user topics,and the users are divided into K clusters with similar interest themes.When personalized recommendation,only similar users in the same cluster are searched,and the traversal range is shortened.Reduce the dimensionality of the data.(4)Considering the similarity and praise of Weibo theme in personalized recommendation,the traditional personalized recommendation only considers the similarity between Weibo themes.This paper combines the similarity of Weibo theme with the amount of praise.The recommendation index is to recommend the Weibo content with high interest recommendation index to the user.Finally,different reference experiments are set up with Weibo user data as the carrier to verify the effectiveness of the proposed method.