Can existing community-detection methods handle bipartite signed networks?

Determine whether community detection methods developed separately for bipartite networks and for signed networks can handle the complexity of bipartite signed networks, in which two distinct node sets are connected by positive and negative links.

Background

Bipartite signed networks involve interactions between two distinct node sets (e.g., users and items) with edges that can be positive or negative. While there is an extensive literature on community detection for bipartite networks and for signed networks separately, the joint setting presents unique structural challenges, such as the absence of triangles (which complicates structural balance considerations) and potential information loss when projecting to one-mode networks.

This paper motivates a systematic evaluation by noting that it is not clear whether established methods built for only one of these dimensions (bipartite or signed) can cope with their combination. The authors subsequently assess algorithms like SPONGE and community-spinglass, but the initial uncertainty highlights a broader methodological question about suitability and robustness in the bipartite signed context.