Combining fuzzy logic and eigenvector centrality measure in social network analysis

•A combined algorithm of fuzzy inference system and eigenvector centrality is proposed.•Social interactions are measured by different factors with different weights.•The influencing factors in a social network are used to weight the friendship strength using the fuzzy logic.•The most influential person is calculated using eigenvector centrality after feeding it with the fuzzy logic results.•The method is applied on large data sets such as Facebook, Epinions, and Slashdot-zoo website.

The rapid growth of social networks use has made a great platform to present different services, increasing beneficiary of services and business profit. Therefore considering different levels of member activities in these networks, finding highly active members who can have the influence on the choice and the role of other members of the community is one the most important and challenging issues in recent years. These nodes that usually have a high number of relations with a lot of quality interactions are called influential nodes. There are various types of methods and measures presented to find these nodes. Among all the measures, centrality is the one that identifies various types of influential nodes in a network. Here we define four different factors which affect the strength of a relationship. A fuzzy inference system calculates the strength of each relation, creates a crisp matrix in which the corresponding elements identify the strength of each relation, and using this matrix eigenvector measure calculates the most influential node. Applying our suggested method resulted in choosing a more realistic central node with consideration of the strength of all friendships.