Research And Implementation Of Microblog Recommendation System Based On MPSO-kmeans Algorithm

With the rapid development of science and technology,microblog has gradually become an indispensable social media in people's daily life because of its fast release,diverse forms and rich content.However,with the explosive growth of information,it is difficult for people to find what they are interested in in a large amount of complex data,so personalized recommendation of blog articles that users are interested in becomes particularly important.Firstly,this paper introduces the related technologies of data acquisition and preprocessing.Secondly,it studies and implements the microblog recommendation system from two parts: discovering topics of interest to users and building influence recommendation model.The research work of this paper mainly includes the following three parts:Firstly,user interest discovery technology is used to obtain topics of interest to users,and k-means algorithm is mainly used to mine interest from microblog data at home and abroad,but the traditional k-means algorithm is vulnerable to the influence of the initial clustering center when mining data.This paper introduces the improved particle swarm optimization algorithm to optimize it,and proposes a particle swarm optimization clustering algorithm(MPSO-kmeans)with dynamic adjustment of learning factor and time factor with weight.MPSO-kmeans algorithm not only overcomes the shortcomings of traditional k-means,but also solves the mutual weakening problem caused by the independent adjustment of inertia weight and learning factor in particle swarm optimization algorithm,realizes the dynamic balance between global search and local precision search,effectively improves the clustering effect,and realizes the efficient mining of topics of interest to microblog users.Secondly,this paper searches blog articles on related topics according to the topics of interest to users,and then obtains the relevant attributes of blog articles and the personality characteristics of their publishers.The principal component analysis method is used to analyze the main components that can fully express the user influence and blog influence.The principal component is used to construct the user influence model and the blog influence model.Combining the two models,a comprehensive influence model is constructed.Finally,the comprehensive influence model is analyzed experimentally,and it is concluded that the model has a high recall rate and accuracy rate for scoring recommendation.Finally,according to the theoretical research and experimental analysis of the above two parts,a system with personalized recommendation function of microblog is designed and implemented.The system can discover topics of interest to users,recommend high-quality blog for users and realize personalized recommendation by sorting the blog searched by the comprehensive influence model according to the influence.