Advances in Complex Systems

Decision-making processes often involve voting. Human interactions with exogenous entities such as legislations or products can be effectively modeled as two-mode (bipartite) signed networks — where people can either vote positively, negatively, or abstain from voting on the entities. Detecting communities in such networks could help us understand underlying properties: for example ideological camps or consumer preferences. While community detection is an established practice separately for bipartite and signed networks, it remains largely unexplored in the case of bipartite signed networks. In this paper, we systematically evaluate the efficacy of community detection methods on projected bipartite signed networks using a synthetic benchmark and real-world datasets. Our findings reveal that when no communities are present in the data, these methods often recover spurious user communities. When communities are present, the algorithms exhibit promising performance, although their performance is highly susceptible to parameter choice. This indicates that researchers using community detection methods in the context of bipartite signed networks should not take the communities found at face value: it is essential to assess the robustness of parameter choices or perform domain-specific external validation.

Heterogeneous networks have multiple types of nodes and edges. Single-layer stochastic block model (SBM), bipartite SBM, and multiplex SBM have been proposed as a tool for detecting community structure in networks and generating synthetic networks for use as benchmarks. Yet, any SBM has not been introduced specifically for detecting community in heterogeneous networks. In this paper, we introduce heterogeneous multilayer SBMs for detecting communities in heterogeneous networks. According to these models, we look at heterogeneous networks as multilayer networks, which means each edge type shows one layer. We can categorize these models into two broad groups, those based on the independent degree principle and other based on the shared degree principle. According to our results, in general, the independent degree model has better performance in networks that have less common communities between nodes types. In contrast, the shared degree model has better performance in networks which have more common communities between nodes types. Also, we show that our models outperform in real-world networks. If we put aside the exception case, simulation results and real data applications show the effectiveness of these proposed models compared to single-layer models that are applied to heterogeneous networks.

The issue of network community detection has been extensively studied across many fields. Most community detection methods assume that nodes belong to only one community. However, in many cases, nodes can belong to multiple communities simultaneously. This paper presents two overlapping network community detection algorithms that build on the two-step approach, using the extended modularity and cosine function. The applicability of our algorithms extends to both undirected and directed graph structures. To demonstrate the feasibility and effectiveness of these algorithms, we conducted experiments using real data.