Detection of Communities on Social Networks Based on Label Propagation Algorithm and Fuzzy Methods

The proliferation of the web and social networks has made people more connected to their friends and neighbors than ever before. The desire of individuals to relate to similar tastes and choices in a social network leads to the formation of clusters or virtual communities. Such information can be useful for commercial, educational or developmental purposes and therefore a large number of algorithms for detecting communities have been presented. There are many algorithms for detecting communities on social networks. In this paper, using the label propagation algorithm and fuzzy Delphi method, an improved method is presented that can identify communities more accurately and quickly than other similar methods. Accordingly, in the proposed algorithm, instead of randomly selecting from the maximum labels of the neighboring nodes, the label with the highest weight is chosen. By doing this, random selection is eliminated, and stability and certainty in the outcomes of the algorithm are achieved.