Combining Tag Correlation and Interactive Behaviors for Community Discovery. (29th September 2018)

• Record Type: Journal Article
• Authors: Li, Chunlin; Bai, Jingpan; Du, Shaofeng; Yang, Chunguang; Luo, Youlong
• Is Part Of: Computer journal. Volume 62:Number 5(2019)
• Page Start: 785
• Page End: 800
• Publication Date: 2018-09-29
• Subjects: community discovery -- Sina microblog -- social network -- Hadoop platform; Computers -- Periodicals; 005.1
• DOI: 10.1093/comjnl/bxy099
• Languages: English
• ISSNs: 0010-4620

Abstract

Abstract: In recent years, Sina microblog becomes an important social network platform for users to not only form personal website but share, deliver and keep interest messages via the relationships between users. However, with larger number of users in social networks, the problem of 'information overload' has been increasingly highlighted. In this case, community discovery has been the urgent requirements in microblog social networks. In this paper, a microblog community discovery algorithm based on tag correlation and interactive behaviors is presented. For the selection of exact user tags, the model of maximal marginal relevance is used and improved to enrich the variousness and novelty of user interest tags. The method of edge-similarity computation is improved to suit for microblog community discovery. Hadoop platform is applied for data process, which greatly decreases the running time of experiments. Finally, extensive experiments are provided to demonstrate the performance of proposed algorithm. The results indicate that the performance of proposed algorithm is better than that of benchmark algorithms for microblog social networks. The proposed algorithm for community discovery can identify the users belonging to multi communities or no community so that the results of community discovery are more accurate and more suitable for actual environment.