Microblog User Influence Based On Improved HITS Algorithm

As one of the most popular social networking platforms,the microblog platform has gradually become a mass media due to its short content and fast information dissemination.With the continuous deepening of influence research,the role of microblog user influence in public opinion dissemination and information guidance has also attracted more and more scholars' attention.Discovering and analyzing high-influence users in the network can accurately grasp the network dynamics in time and guide the trend of network public opinion.Accurately calculating the influence of microblog network users has become a hot issue in this field.Based on the in-depth analysis and measurement of microblog user relationship,this paper proposes a microblog user influence calculation method based on improved HITS algorithm,considering the individual differences of influence from three aspects of network structure,interaction behavior and time,aiming at the problems of unclear user relationship and unobvious time characteristics in traditional microblog network user influence measurement methods.This method takes the dynamic relationship between users as the breakthrough point,selects the dynamic relationship strength as the edge weight of the improved algorithm,modifies the center value through sigmoid function,integrates the authority value and the center value of users to quantify the user influence,improves the adaptability of the algorithm,and the calculation results are more reasonable,which provides a new idea for the calculation of microblog user influence.The main work of this paper includes two aspects:(1)Calculate the strength of microblog user relationship.Analyzed the influencing factors of the user relationship strength,and proposed a microblog user relationship strength calculation model.This method is based on user attribute characteristics and user interaction behavior,and mines the strength of the relationship among microblog users from three dimensions of user similarity,trust,and interaction.Experimental results on real data show that this method can effectively calculate the strength of the relationship between users.Compared with traditional user relationship calculation methods,the model has a significant improvement in NDCG value and accuracy.(2)Calculate the influence of microblog users.Firstly,an improved HITS algorithm is proposed,which integrates the relationship strength between microblog users on the basis of the original algorithm,and improves the adaptability of the algorithm in microblog network user influence analysis;secondly,considering that the existing methods do not fully consider the dynamic relationship characteristics of users,the time factor is added to microblog network on the basis of integrating the relationship strength of users In the construction of the network,the network data is divided according to the unit time,and the user interaction strength in each time cycle is calculated.Combined with the improved HITS algorithm,the user influence is calculated.Taking Sina Weibo as the experimental object,the experimental results show that,compared with the traditional influence calculation method,this method has improved in DCG value,precision rate,recall rate and F-score evaluation index.