Multiple Contrastive Constraints Based Deep Nonnegative Matrix Factorization For Community Detection

Community detection is a fundamental task in complex network analysis, aiming to identify closely connected subgraphs within the network. Recently, nonnegative matrix factorization (NMF)-based community detection methods have garnered significant attention due to their excellent clustering performance. However, most existing approaches are shallow and fail to fully capture the neighborhood information of nodes, which plays a critical role in node embedding learning. Considering the recent success of graph contrastive learning, its explicit structural constraint characteristics can effectively guide the model to learn embeddings useful for node clustering. To address the problems mentioned above, we propose a deep nonnegative matrix factorization model based on multiple contrastive constraints (CCDNMF). By incorporating the multi-neighborhood sampling method based on multiple constraints contrastive learning, this approach leverages node neighborhood features from different perspectives, capturing a rich set of local data distribution characteristics. This enables the learning of embeddings that preserve both global semantics and hierarchical local similarities, effectively integrating DNMF’s interpretability with the representational strengths of contrastive learning, and enhancing community discovery in complex networks. Extensive experiments are conducted on eleven real-world datasets that demonstrate the effectiveness and rationality of the proposed approach. The experimental results show that CCDNMF is superior to most existing methods. Supplementary Material File (ccdnmf.pdf) - Download - 934.49 KB Information & Authors Information Version history Copyright This work is licensed under a Non Exclusive No Reuse License.

Authors Metrics & Citations Metrics Article Usage 182views 120downloads Citations Download citation Langcai Cao, Huan Liu, Bilian Chen. Multiple Contrastive Constraints Based Deep Nonnegative Matrix Factorization For Community Detection. Authorea. 24 August 2025. DOI: https://doi.org/10.22541/au.175605324.43116092/v1 DOI: https://doi.org/10.22541/au.175605324.43116092/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu.