Research On Personalized Microblog User Recommendation Algorithm Based On Multi-source Information

With the rapid development of information technology and social network,more and more people participate in social networks.As an important information interaction platform,micro-blog has the characteristics of sharing,timeliness and interactivity,which is quickly favored by a large number of users.Naturally,the question about how to quickly and accurately find the user's interest in a large number of users of micro-blog becomes a hot issue in the current research.In this thesis,the classification efficiency of the KNN algorithm decreases with the increase in the size of training set and feature dimension,a novel text classification algorithm is proposed.In order to make up for the defect existing in traditional collaborative filtering recommendation algorithm such as data sparsity and low accuracy rate,this thesis proposes a micro-blog user recommendation algorithm based on the similarity of multi-source information.The main work of this paper includes the following two aspects:(1)A CRS-KNN text classification algorithm is proposed based on Canopy and Rough Set.Firstly,the text data is processed by a clustered algorithm called Canopy.Then every cluster is segmented by the theory of upper and lower approximation of Rough Set.It's worth noting that the lower approximate region without classification.The final category is determined through the KNN algorithm for the data of boundary from lower and upper approximation difference.This method not only reduces the data calculation scale of KNN algorithm,but also improves the classification efficiency,and the accuracy rate,recall rate and F1 value are improved.(2)This thesis proposes a micro-blog user recommendation algorithm based on the similarity of multi-source information.Firstly,the users are classified by CRS-KNN algorithm according to the micro-blog user label information;then calculate the similarity for multi-source information in each class of user.Secondly,the total similarity calculation of multi-source information is introduced between the time weight and richness weight,and recommended the TOP-N users which had larger value to users.Finally,a parallel computing framework Spark is built for verifying the validity of the algorithm.The experimental results show that the proposed method is not only effective in user recommendation,but also has a significant improvement in accuracy,recall and efficiency.This thesis analyses the individualized characteristics of micro-blog users.According to the classification algorithm,the multi-source information of similarity computation and influencing factors of total similarity about micro-blog user,this thesis puts forward a multi-source user recommendation algorithm based on micro-blog users' classification.It verifies the final results using the data of Sina micro-blog and obtains good effectiveness.The work has a fine reference value for research of micro-blog users personalized recommendation algorithm.