Exploring Uncertainty Methods for Centrality Analysis in Social Networks

Network centrality reflects node importance in networks, which is a challenging problem in social network analysis. Based on Fuzzy Set and MYCIN theory, this paper proposes a novel node centrality measuring method and models n-monkeys dataset, where n is 20. Initially, we created monkeys relationship graph and generated relationship matrix based on the monkeys' encountering times in a specific time period and location, and calculated degree and average distance for each individuals. Then, we performed fuzzy processing on degree, average distance, age and sex, and define authority in the different domains. At last, we modeled centrality using MYCIN combination for the social network node. On the standards dataset WOLFE PRIMATES with 20 monkeys, we evaluated our algorithm and compared it with original rankings in terms of precision, which reached 82.5% for fuzzy set based approach and 76.6% for MYCIN based approach, with 10.9% and 5% improvements over current best practices respectively, indicating that fuzzy set and MYCIN models is reasonable and effective in social network analysis.