The Detection And Control On Abnormal Users Of Online Social Network

In social network,abnormal users are always existent.At present,the detection processes for those abnormal users are generally designed on basement with the feature of information,these detection processes exist some shortcomings as high computation cost,low detection efficiency,bad influence on user experience,and so on.By reason of lacking of systematic and corresponding control strategy,the detection ability and control ability to abnormal users are always limited.Based on the concept of information entropy,this paper proposes a creative solution,that is concise and effective,to risk assessment for abnormal users in online social network.Then this paper chooses and improves a kind of access control model,which is applicable to online social network,with the encourage of information sharing between users,the allowing of a handful of risks,and corresponding risk control processes that could efficient manage abnormal users according to the detection results.At first,this paper introduces the present situation of the related research about the abnormal users in online social network,and analyzes the difficult point on this topic research,and expounds the involved knowledge background in this paper.Then this paper put forward the concept of Interaction Entropy,and describes the related definition and properties.After the declaration of data source,this paper expounds with example in detail that how to determine the detection period,what is the specific steps on the Interaction Entropy calculation on users' message log data.After that,this paper improves the Fuzzy MLS model according to the online social network environment.After mapping the value-at-risk form detection process to the relevant variable in Fuzzy MLS model,this paper further amends on the model,and develops a kind of effective social network anomaly detection and control model.On this basis,this paper uses real data to calculate and statistical analysis the users' Interaction Entropy.And then it maps the result into the Fuzzy MLS model,and completes the model building and the control strategy.With the validation and analysis on test data,the experiment verifies this research.The last is the summarization of the thesis,and makes follow-up research ideas and outlook.