| With the continuous improvement and wide application of Internet-related technologies,online social networks have become the main way for people to share information.As an important information interaction platform in social networks,microblog has been favored by a large number of users.With the increase of microblog users,the user data of social platforms has grown exponentially,and related system applications have also increased.User similarity calculation is an important basis for social platform related applications,such as social recommendation,community discovery,etc.,which has attracted the attention of many researchers.How to mine more similar users in microblog social network more accurately and help the platform to provide users with better services is a hot issue of current research.In this research,we conducted in-depth analysis of microblog users and proposed a method for mining similar users on microblog.First,this paper proposes a microblog important user mining algorithm for the problems of microblog user influence evaluation method.Then,it clusters important users,generates interest expressions for users,and combines user background information to propose a method for calculating the similarity of microblog users with interests.The specific work is as follows:(1)Study the influence analysis of microblog users,and propose a method for mining important users of microblog which improves PageRank.First of all,analyze the background information of the users,define the basic self-influence for each user,and aim at the problem that the users are inactive for a period of time and the influence is misjudged and decreased.Based on the interactive information of the user’s blog posts,a calculation method of blog post propagation rate is introduced.Finally,based on the user’s social relationship,through improving the PageRank algorithm,a user influence evaluation algorithm was constructed to mine important users.The experimental results indicate that the method has improved accuracy and recall,and validate the effectiveness of the proposed method.(2)The interest expression of users is researched,and a similar user computing method of microblogs integrating interests is proposed.This paper analyzes some current researches on microblog user interest mining,and proposes a method of indirectly acquiring user interests based on important users of microblog relationship network.In order to enrich the vectorized representation of user labels,Word2 Vec is used to generate low-dimensional word embedding vectors with semantic information,and the vectors are clustered to obtain the user’s interest expression.Combined with the background similarity of the user,a new method based on user interest and a comprehensive similarity calculation model for background information,which hierarchically mines similar users on microblog.Experiments are performed on actual datasets,and the results show the effectiveness of the proposed algorithm.In summary,this paper analyzes the relevant characteristics of microblog users,conducts in-depth research on user influence evaluation,user interest mining,and finally proposes a method for mining similar users of microblogs with combined interests.The experimental results show that this method has a good effect on mining similar users on microblog,which has an important role in the fields of user recommendation,community discovery,and influence analysis. |
