Research On Analyzing Method Of Collective Attention In Social Network

Collective Attention refers to some attention to one same thing formed by group users.Collective Attention is one of the important research areas in social network. Recent studyabout collective attention mainly focuses on analyzing the tendency of attention aims atmaking the researchers or site operators accurately discern and predict the attention tendency.By analyzing the popularity of events, we can correctly grasp the core content of socialnetwork or business site, find out the event that collective users concern about at some time,and help users to screen out the favorite and most popular topic or event in time. How to makemodel about collective attention, and how to classify the attention and predict the attentiontendency according to the information or content of social network is the major problemneeded to be solved, it’s also the main research content of our paper. The main work of ourpaper is as follows:Firstly, by analyzing the existing research and existing problems of collective attention,we propose a model of collective attention: peak-based collective attention model, which canbe short for PebCAM. PebCAM is able to describe the variation tendency of collectiveattention along with time, and can give the computational formula of attention peak value.PebCAM provides data base for subsequent dynamic classification of collective attention.Secondly, based on PebCAM model, we study the dynamic classification of collectiveattention CACM. CACM method can divide the collective attention along with the passage oftime and the change characteristics of peak, according to the way that collective attentionepidemic characteristics reach to peak and the different changes of peak between before andafter, collective attention can be classified into continuous before peak, continuous after peak,continuous symmetrically and continuous singly four categories. Through the case study andexperimental verification, the correctness and feasibility of the dynamic clustering method isproved.Finally, we propose the CASPM algorithm to predict the collective attention. TheCASPM algorithm is based on PebCAM model, it uses the propagation quantity of topic topresent collective attention, by analyzing the initial propagation quantity and the novelty oftopic to predict the future propagation quantity of some time, so that we can get the future tendency of collective attention and be applied to predict the popularity of event. Theextensive experiment is performed to verify and analyze the performance of proposedCASPM algorithm. By comparison of different prediction algorithms, the experiment verifiesthe feasibility of the proposed CASPM algorithm.