Research On The Influence Of Micro-blog Users Based On Relational Analysis

With the development of social networks,social applications have become an important platform for information and communication sharing.The information released by users in social applications involves many fields,which has a certain impact on various industries.Therefore,it is of great significance to evaluate the influence of the user in information dissemination and business marketing.At present,many domestic and foreign researchers have studied the influence of social network users,but the processing of massive data and the analysis of complex network structure are the key to the research of influence.In view of the above problems,based on the user relationship analysis,this paper studies the influence of micro-blog users from two layers of communication influence and structural influence.The main contents of this paper are as follows.In the aspect of user communication influence,aiming at the problem of user data processing,a ranking algorithm based on active forwarding relation is proposed to filter non influential users in the network.The algorithm uses a random forest classification model to predict the activation and forwarding state between users.It then activates and filters users in the network according to the principle of the independent cascading model,and constructs a new activation network.Finally,it calculates the activation ability of the users in the network and rank according to the user activation ability.The experiment shows that the algorithm proposed in this paper can reduce the time of the ranking of the influence users and improve the efficiency of ranking compared with the correlation ranking algorithm.In terms of structure impact,a modified PageRank algorithm based on user relationship is proposed for complex network structure problems,which improves the weight partition problem of PageRank algorithm.The algorithm uses the machine learning support vector regression algorithm to predict the user relationship.Then according to the tightness of the user relationship,the weight of the influence transfer between the users in the transfer matrix is improved,and finally the ranking is made according to the value of the user's influence.Experiments show that compared with the related influence ranking algorithm,the algorithm proposed in this paper has more advantages in influencing user's forwarding volume,commenting volume,praising volume and topic volume.The influence comprehensive analysis system is designed and implemented,and the scoring strategy is used to rank the comprehensive influence of the users.The system has realized the ranking of the communication influence,the structure influence and the comprehensive influence of the user,which has the advantages of simple and easy to use and good performance.