Research On Interest Evolution Of Online Users

Users' clicks and jumps in the network can be regarded as a kind of movement in virtual space,which is usually driven by human interests.Though many breakthroughs have been made in exploring the dynamics of human behavior,little is known about the dynamic evolution of human interest in virtual space,partly due to the extreme difficulty in getting enough behavioral data and accessing the human mind from observations.Based on the massive online user behavior data provided by China Internet Information Center,this paper regards every website that users click on online as a point of interest,and studies the evolution of user interest in virtual space.The main research work is summarized as follows:Firstly,To solve the problem that online users' interest is difficult to quantify,this paper proposes an interest stickiness measurement algorithm.The study makes an empirical analysis of online users' interest behavior from the aspects of continuous moving step length,return step length,dwell time and so on,and reveals the scale-free feature of power law in human online behavior.Based on the analysis results,an interest stickiness measurement algorithm to measure the attraction of points of interest to users is proposed.Through the experimental comparison with PageRank algorithm and attention flow model,the effectiveness of variables and algorithms is verified.Secondly,In order to measure the similarity of interest points in the network space and study the jump regular pattern between interest points,this paper proposes a similarity algorithm of interest nodes based on spatial position.Through the optimization of SPA model,the two variables of node generation time and radius of influence are deduced from the node's in-degree,and the behavioral variable of interest stickiness is introduced,so that the interest nodes in the network can be represented by these three variables.The spatial norm is used to calculate the distance between node pairs in the clickstream network.The experimental results show that the nodes with high in-degree are more attractive,and the smaller the distance between pairs of interest nodes is,the more similar the interest nodes are.Finally,The fluidity evolution model of online user interest is constructed.The model consists of two complementary behavior variables: exploration and preferential return.These two important behavior mechanisms promote the interest transfer of online users.The experimental results indicate that with the increase of the number of interests,the possibility of online users to explore new interests decreases exponentially,and users show a stronger return preference for the interest with high frequency of previous visits.