Community Detection for Heterogeneous Multiple Social Networks

TL;DR

This work tackles cross-network community detection across heterogeneous social networks by learning a global fused community via a common consensus matrix $C$ within a nonnegative matrix tri-factorization framework. It jointly models topology and content, incorporates overlapping users through alignment matrices $H^s$ and $T^s$, and enforces a global fusion constraint to produce a coherent community structure across networks. Convergence of the optimization is established, and the approach is validated on synthetic benchmarks and real data (Twitter, Instagram, Tumblr), showing superior community quality and fusion as measured by modularity, compactness, density, and NMI. The results demonstrate HMCD’s ability to leverage overlapping users to synthesize a global community, enabling insights into information diffusion and behavior migration across platforms.

Abstract

The community plays a crucial role in understanding user behavior and network characteristics in social networks. Some users can use multiple social networks at once for a variety of objectives. These users are called overlapping users who bridge different social networks. Detecting communities across multiple social networks is vital for interaction mining, information diffusion, and behavior migration analysis among networks. This paper presents a community detection method based on nonnegative matrix tri-factorization for multiple heterogeneous social networks, which formulates a common consensus matrix to represent the global fused community. Specifically, the proposed method involves creating adjacency matrices based on network structure and content similarity, followed by alignment matrices which distinguish overlapping users in different social networks. With the generated alignment matrices, the method could enhance the fusion degree of the global community by detecting overlapping user communities across networks. The effectiveness of the proposed method is evaluated with new metrics on Twitter, Instagram, and Tumblr datasets. The results of the experiments demonstrate its superior performance in terms of community quality and community fusion.

Figures (10)

• Figure 1: Community structure across social networks. In this figure, users A, B, C, D, E, F and G are overlapping users from the three social networks, OSN1, OSN2 and OSN3. These users have different community structures in their social networks respectively. The main difficulty in implementing community detection across multiple networks is integrating the communities of diverse networks with overlapping members as the core. In OSN1, user C and user F are in the same community, and in OSN3, user F and user D are in the same community. Users C and D can be grouped in the same community by analyzing transitive relation. This division corresponds to the community division of OSN2's users C and D. However, in another case, users A and B may be part of the same community in OSN3 but belong to different communities in OSN1.
• Figure 2: Research framework
• Figure 3: Illustration of the proposed community detection model across multiple social networks. Our approach primarily takes into account some overlapping users across multiple social networks. Therefore these networks are partially aligned. We posit that the cohort of overlapping users forms the essence of community across these networks, with the community forged among these users potentially exhibiting significant disparities from those among regular users. Furthermore, the community of overlapping users serves as a pivotal component for fostering community fusion.
• Figure 4: In alignment matrix operations, the value of an element in the matrix indicates whether the row user $u_i$ and the column user $o_j$ are the same users.
• Figure 5: Example of the photo posted