Research On Microblog User Behavior And Information Dissemination

As Internet is developing,microblog has developed into an important platform to communicate and share information.Microblog users promote the spread of information through releasing,paying attention,forwarding and making comments.Microblog users’ own behaviors not only affect the spread of news,but also affect the direction of news propagation.It has a great significance for fast detection of information dissemination,avoid malicious information caused great influence,timely detect the hot issues or malicious information transmission range and safeguard social stability.Through analyzing the user behavior,studying the background of information transmission as the background,according to the characteristics of microblog high-dimensional sparse,this paper will research the larger influence of k user nodes of acquisition,information transmission path of microblog users behavior characteristics,aims to build the user behavior feature model,thus can get information transmission quickly.The main research is as follows:(1)The conservative approach is taken to extract useful field of microblog,word segmentation,splitting on whitespaces and punctuation mark,and eliminating stop words to obtain pure data.(2)Based on the micro-blog user behavior(release,attention,forwarding and comment)and micro-blog contents,the writer puts forward a kind of user influence algorithm based on user behavior characteristics and micro-blog contents.Through analyzing microblog user behaviors(release,attention,forward and comments,etc.)and get behavior data,then the user influence weights can be calculated.Using microblog user contents to build word co-occurrence matrix,then use LDA model to identify the underlying theme,through KL divergence to obtain similarity between user,finally combine user influence weights to get influence of the customers.Experiments show that, this algorithm is more effective.(3)The user influence algorithm combined with user microblog contents and structure relationship has proposed. Through attention and be concerned by microblog to establish user relationship structure.Combined with the relationship between user microblog contents and contact structure,a new influence algorithm CSIAI that fused user contents and relationship structure has proposed.This method first builds the users’ word document matrix and user relationship structure,through the mutual information between words and similarity between initialization words,and then iterative the similarity between calculate words and user documentation,through weighting the user influence weight,thus get users’ adjacency matrix.Experiments show that k node users with larger influence who extract and observe can be able to detect the spread of microblog information quickly.We focus on the three points above on microblog user influence.Experiments show effectiveness of our method.