Social Network Information Filtering And Recommendation System Implementation Based On Machine Learning

With the rapid development of the Internet,the immediacy and sharing of information transmission have been greatly improved.People's social centers are gradually migrating from offline to online.Online social networks have become a main field of social network because of their vast user groups and open information dissemination.Every day,the data generated on social network is up to the PB level,and the excessive data information causes the social network to face the problem of noise information and information overload.There is some useless information on the social network,and the user can't find the interested content irmmediately,which seriously affects the user's social experience.In response to this problem,this article carries on the solution from two aspects.On the one hand,this paper studies the related technologies of information filtering system for the problem of low information quality.The information filtering method based on machine learning has the advantages of high accuracy and fast speed.However,these methods only use surface word features such as mutual information,the text feature extraction method is single.And the methods neglect the diversity of noise information.Aiming at this problem,this paper proposes an improved information filtering algorithm based on machine learning.According to different language usage and distribution forms,the noise information is accurately classified into four categories.Each type uses the classifier model in a targeted manner.Surface word features are combined with deep learning semantic word vector features,riching text feature expression.The experimental results show that the proposed method achieves a better filtering effect on the Douban data.On the other hand,for the problem of information overload,this paper studies the related technologies of information recommendation system.The collaborative filtering recommendation algorithm based on machine learning has good recommendation effect,but these methods only recommends by analysing user score or overall emotion extracted from user comments,regardless of the emotional bias of different attributes in user reviews.Aiming at this problem,this paper proposes an improved machine learning-based information filtering algorithm.The research extracts the attribute words in the user comments through the association rule model,uses the topic model to aggregate the attribute words into attribute faces,and then uses sentiment analysis to calculate the user's emotional tendency on each attribute face.The information is applied to the recommendation system.Considering the text non-standard characteristics in the social information,the steps of judging the emotional level of the unknown vocabulary are added.The experimental results show that the proposed method achieves a better filtering effect on the Douban data.As an improvement of the traditional algorithms,this paper proposes a novel information filtering and recommendation algorithm based on machine learning,as well as designs and implements a social network information filtering and recommendation system,and experiments have good results.Future research will consider the relevance of user reviews to improve the recommendation.