A novel approach for choosing dynamic community detection algorithms using PROMETHEE II

Abstract

In this paper, a decision support tool is developed in order to assist senior managers to choose a dynamic community detection algorithm. The main objective behind this novel approach is to improve the search strategy of algorithms to decrease the search time of the best fitting one according to the study case. The suggested study ranks existing dynamic community detection algorithms in order to find the most appropriate one according to some preferences. For that purpose, we need a multi-criteria ranking method such as PROMETHEE II. This latter allows ranking alternatives from the best to the least according to some criteria extracted from the analysis of a provided questionnaire on dynamic community detection algorithms. The process of our approach is as follows: first, the questionnaire is provided to some experts in community detection domain. Second, answers are collected and analyzed in order to extract criteria. Third, a ranking of the algorithms with PROMETHEE II is operated according some extracted criteria and user preferences. Finally, the first ranked algorithm by PROMETHEE II is provided to the user in order to detect the communities.