The Research Of Social Network Data Mining Based On Collective Intelligence

In a variety of social network,weibo as the representative of the new applications continues to break the traditional information transmission and communication and interpersonal communication model.The spread of unverified information is potentially a problem,hampering the healthy development of social networking environments.The study of social network has important significance in establishing correct public opinion and curbing the spread of rumors.In the past,social network research rarely analyzes the weak relationship,but the weak relationship greatly affects the dissemination of information.On the basis of determining the social network relationship,the social network research has sufficient basis.Ranking research on social networking rumors helps to distinguish between rumors of high activity and inactive rumors,thereby reducing the cost of rumors detection.In the rumor detection,although the ant colony algorithm can improve the accuracy of rumors classification,but the ant colony algorithm is a global optimal algorithm,the efficiency is low,combined with random forest feature selection,can make up for this defect.Therefore,this thesis mainly focuses on the analysis of social network relations,weibo rumor rankings,weibo feature analysis and rumor detection in three aspects of in-depth study.The main work of this thesis includes: the research of weibo data collection and preprocessing technology.In view of the specific groups in social network,the relationship between rumor and non-rumor users,the influence of various factors were concerned on the ranking of rumors in the ranking algorithm.On the foundation,weibo rumor ranking algorithm was proposed.After the preliminary analysis of the weibo data,the paper classified on features of user,propagation.Based on classic algorithm,ant colony and random forest Mixed feature selection algorithm was proposed,experiments were done on the algorithm accuracy and recall rate,the results show that the improved algorithm made the rumor detection accuracy better.