A novel intelligent Fuzzy-AHP based evolutionary algorithm for detecting communities in complex networks

The realm of complex network analysis is witnessing a surge in research focus on community detection. Numerous algorithms have been put forth, each harboring distinct advantages and drawbacks. Predominantly, these algorithms rely solely on network topologies for community detection. Yet, many real-world networks harbor valuable node content that intricately mirrors the fabric of their communities. Recognizing this, leveraging node contents stands as a potential avenue to augment the quality of community detection. This study introduces an innovative evolutionary algorithm rooted in the fuzzy analytical hierarchy process (FAHP) to propel community detection in complex networks by intertwining content and structural information. Noteworthy is its departure from the conventional multi-chromosome evolutionary algorithms, opting for a single-chromosome design that substantially curtails computational complexity. The algorithm employs a distinctive FAHP-based local operator, termed the community topological modifier, to refine community structures and elevate the quality of community detection within the current generation. A novel criterion for gauging content similarity among nodes is integrated into the algorithm. Additionally, an early fusion approach is suggested, creating a hybrid graph that amalgamates structural and content information between nodes. Rigorous evaluation in diverse real networks ensued, with comparative analyses against state-of-the-art and traditional methods. Notably, the proposed algorithm emerged as the frontrunner, securing top rankings across all evaluation criteria—such as normalized mutual information (NMI) and adjusted Rand index (ARI)—based on the results of the Friedman test.