In social networks, community structure arises from a complex interplay between structural features and actor attributes. This study aims to enhance density-based community detection by integrating both topological and attribute information within a unified framework. Evaluations on simulated and real-world datasets demonstrate that incorporating attribute information improves detection performance in networks with moderate to high mixing. Results suggest that the optimal combination of topological and attribute information depends on the network structure and the type of attribute data, with a balanced approach often yielding the best results. The study also highlights the importance of choosing the topology similarity measure, with the neighbor set similarity approach proving to be more robust.
