Improving social network community detection using DBSCAN algorithm

Social networks depict the interactions between individuals or entities and are represented by a graph of interconnected nodes. The study of such graphs leads to understanding of this data and concluding different communities. Among the different clustering algorithms, DBSCAN is an effective unsupervised clustering algorithm which is implemented in this work to emphasize community detection in social network. The results specifies the number of high influence members represented by core, less influence represented by border and members with no influence in the groups represented by outliers. By eliminating the outliers the dataset will be noise free to deal with it.