Modularity in Multilayer Networks using Redundancy-based Resolution and Projection-based Inter-Layer Coupling

Alessia Amelio, Giuseppe Mangioni, Andrea Tagarelli

Abstract

The generalized version of modularity for multilayer networks, a.k.a. multislice modularity, is characterized by two model parameters, namely resolution factor and inter-layer coupling factor. The former corresponds to a notion of layer-specific relevance, whereas the inter-layer coupling factor represents the strength of node connections across the network layers. Despite the potential of this approach, the setting of both parameters can be arbitrarily selected, without considering specific characteristics from the topology of the multilayer network as well as from an available community structure. Also, the multislice modularity is not designed to explicitly model order relations over the layers, which is of prior importance for dynamic networks. This paper aims to overcome the main limitations of the multislice modularity by introducing a new formulation of modularity for multilayer networks. We revise the role and semantics of both the resolution and inter-layer coupling factors based on information available from the within-layer and inter-layer structures of the multilayer communities. Also, our proposed multilayer modularity is general enough to consider orderings of network layers and their constraints on layer coupling. Experiments were carried out on synthetic and real-world multilayer networks using state-of-the-art approaches for multilayer community detection. The obtained results have shown the meaningfulness of the proposed modularity, revealing the effects of different combinations of the resolution and inter-layer coupling functions. This work also represents a starting point for the development of new optimization methods for community detection in multilayer networks.

Introduction

In network science, modularity is the standard metric for evaluating how well a network is partitioned into communities. While well-defined for single-layer graphs, extending this to multilayer networks—where nodes interact across multiple types of relations—introduces significant complexity.

The current standard, multislice modularity ($Q_{ms}$), relies on two parameters: a resolution factor ($\gamma$) and an inter-layer coupling factor ($\omega$). Traditionally, these parameters are set as global constants. However, this approach ignores the unique topological characteristics of individual layers and communities. Furthermore, $Q_{ms}$ lacks native support for "ordered" layers, such as temporal snapshots of a network, where coupling should ideally respect the chronological sequence. This paper proposes a more principled, data-driven approach to multilayer modularity that eliminates arbitrary parameter tuning.

Methodology

The authors introduce a new formulation of multilayer modularity ($Q$) that replaces static parameters with dynamic, structure-aware factors.

1. Redundancy-based Resolution ($\gamma$)

Instead of a fixed global value, the resolution factor is calculated for each community-layer pair. It leverages the concept of "redundancy"—the variety of link types between node pairs. In communities with high information content (many redundant links), the resolution factor is automatically adjusted to reflect the layer's specific contribution to the internal community structure.

2. Projection-based Inter-Layer Coupling (IC)

The coupling strength is determined by the overlap between a community's "projection" (its node set) across different layers. This replaces the constant $\omega$ with a measure of structural similarity.

3. Layer Ordering and Asymmetry

The model explicitly supports partial order relations, allowing for different coupling strategies depending on the network type. For instance, in temporal networks, layers can be coupled only to their immediate successors (adjacent coupling) or to all future layers (succeeding-layer coupling).

Illustration of multilayer network structure with adjacent and succeeding-layer coupling.

Additionally, the framework introduces asymmetric coupling. This recognizes that the relevance of a community might differ when projecting from a "source" layer to a "destination" layer, capturing the intuition that a small, tight community projecting into a larger one is more significant than the reverse.

Experimental Evaluation

The proposed metric was validated using both synthetic benchmarks and diverse real-world datasets, including social networks (Twitter, YouTube), professional relationships (AUCS), and infrastructure networks (EU-Air, London transport).

Key Results

  • Enhanced Expressiveness: The proposed $Q$ consistently outperformed the standard $Q_{ms}$ in differentiating community structures across synthetic benchmarks.
  • Information Capture: Using redundancy-based $\gamma$ led to modularity gains of up to 0.12 compared to fixed $\gamma=1$, demonstrating that data-driven resolution captures more structural information.
  • Directional Relevance: Asymmetric coupling ($IC_a$) generally yielded higher modularity scores, suggesting that directional dependencies are a fundamental feature of multilayer systems.
  • Topological Correlation: A strong positive correlation was found between the new modularity measure and edge-set coverage, confirming its reliability as a proxy for community quality.

Discussion and Future Work

By making resolution and coupling community-specific and time-aware, this research moves away from "one-size-fits-all" constants toward a more nuanced understanding of multilayer structures.

However, this increased precision comes with a computational cost; calculating variable factors increased processing time by approximately 50% in tests. Future work will focus on optimizing these calculations for large-scale multiplex graphs. Additionally, while the current model handles "crisp" community memberships, extending the framework to accommodate overlapping communities and developing dedicated optimization algorithms to maximize this new $Q$ are key next steps.

Reference: Amelio, A., Mangioni, G., & Tagarelli, A. (2019). Modularity in Multilayer Networks using Redundancy-based Resolution and Projection-based Inter-Layer Coupling. IEEE Transactions on Network Science and Engineering. DOI: 10.1109/TNSE.2019.2913325.

Summary updated

Abstract

The generalized version of modularity for multilayer networks, a.k.a. multislice modularity, is characterized by two model parameters, namely resolution factor and inter-layer coupling factor. The former corresponds to a notion of layer-specific relevance, whereas the inter-layer coupling factor represents the strength of node connections across the network layers. Despite the potential of this approach, the setting of both parameters can be arbitrarily selected, without considering specific characteristics from the topology of the multilayer network as well as from an available community structure. Also, the multislice modularity is not designed to explicitly model order relations over the layers, which is of prior importance for dynamic networks. This paper aims to overcome the main limitations of the multislice modularity by introducing a new formulation of modularity for multilayer networks. We revise the role and semantics of both the resolution and inter-layer coupling factors based on information available from the within-layer and inter-layer structures of the multilayer communities. Also, our proposed multilayer modularity is general enough to consider orderings of network layers and their constraints on layer coupling. Experiments were carried out on synthetic and real-world multilayer networks using state-of-the-art approaches for multilayer community detection. The obtained results have shown the meaningfulness of the proposed modularity, revealing the effects of different combinations of the resolution and inter-layer coupling functions. This work also represents a starting point for the development of new optimization methods for community detection in multilayer networks.
BibTeX
@article{amelio2020modularity,
  author = {Alessia Amelio and Giuseppe Mangioni and Andrea Tagarelli},
  title = {{Modularity in Multilayer Networks using Redundancy-based Resolution and Projection-based Inter-Layer Coupling}},
  journal = {IEEE Trans. Network Science \& Engineering},
  year = {2020},
  url = {https://openreview.net/forum?id=Z2VIvCDsDnW},
}
Figure