Abnormal Social Network User Identification System Design And Implementation Based On Attribute Reduction

Microblog has become an important part of social media,a large number of users send and thus spread information on this platform.Nowadays,the network environment of microblog is impacted by the presence of anomalies in users seriously.So the research on identifying the types of microblog users is of great significance.Based on the example of microblog,this paper selects some microblog users as research objects and thus analyzes and extracts the features of the selected users,then selects the features of users through attribute reduction.Meanwhile,it uses statistical methods and classification methods in data mining to analyze user data.With the breakthrough point,the classification method C4.5 Decision Tree,and compared with several other classification methods,this paper has trained the history data to form a classifier to make prognostic classification of new sample,which has realized high accuracy.Finally based on C4.5 Decision Tree classifier,the attribute reduction is also added to the research.It achieves the pruning of decision tree and thus further improves the accuracy of the predictions,it can make classification accuracy reach rate of 92.68%.Not only based on the research of Weibo user characteristics,but also based on the micro-blog content for text classification using Naive Bayesian classification methods.After the Chinese Word Segmentation,Chinese word frequency statistics and go to stop word of text,microblog can be represented by a vector space model,and finally realized using weka Naive Bayes classifier experiments.Since the stage consider of going to stop words,in addition to consider microblog disable features itself,the classification accuracy rate can reach 88.65%,and achieved good classification results.Finally,based on the theory of microblog users classification and microblog text classification,design and implement a Weibo user recognition system.Not only can analyze data of Weibo user,but also process batch of Weibo user data to judge their user types,having an important practical significance.