Research On Analyzing Influence Of Users In Campus Forum

At present,the analysis of social network users' influence and influence ranking has become an important work in the field of social network research.As an online social platform with long history and mature technology,web forum has a large number of users.In this paper,we take the campus forum as the research object,and study the algorithm of finding opinion leaders in campus forum,which has the practical significance of guiding the information dissemination and building a harmonious campus.The existing methods generally analyze the influence based on social network structure and user attributes,without considering the attenuation of the user influence on the time dimension.In order to accurately identify the high-impact users in the forum,this paper first build a multi version dataset based on different time points,and establish user interaction network model and conduct user behavior analysis.Compared with the existing literature only use a single time point dataset for analysis,in this paper,a multi version dataset is constructed in the form of total quantity and increment portion.By comparing the data differences between different versions to make the behavior changes analysis,we find the behavior of users follow the power law distribution.Meanwhile,this paper constructs a user interaction network model based on the theme,and the number of user interaction is the edge power,through the data collected,the results show that the model can reflect the difference of influence between different users.Based on the above work and the existing problem of PageRank algorithm,we propose Time-User Rank algorithm.This algorithm combines time factor and considers the changing characteristics of influence.We fits the influence attenuation coefficient formula based on the changes of the probability of response to the topic post over time and introduces it into the algorithm.At the same time,this paper proposes to differentiate different users by using the weight of the interaction between users and degree of user topic participation,and optimize the weight distribution of nodes.Finally,this paper verifies the effectiveness of the Time-User Rank algorithm.By comparing with the classical PageRank algorithm,Time-User Rank can identify high-impact users with more central features and higher user coverage in a single time period.In continuous time period,due to the introduction of influence attenuation coefficient,the users identified by Time-User Rank algorithm have relatively higher and more stable user coverage,so as to achieve better recognition results.