[Paper Review] A New Community Definition For MultiLayer Networks And A Novel Approach For Its Efficient Computation

Abhishek Santra, Kanthi Sannappa Komar+2 more|arXiv (Cornell University)|Apr 20, 2020

Complex Network Analysis Techniques36 references4 citations

TL;DR

This paper introduces a novel structure- and semantics-preserving community definition for heterogeneous multilayer networks (HeMLNs), enabling accurate and efficient detection of multi-layer communities without information loss. Using a decoupled computation framework based on bipartite graph pairings and customizable weight metrics, the approach achieves high efficiency—iterative costs are an order of magnitude lower than initial one-time costs—while supporting flexible analysis across diverse objectives, as validated on IMDb and DBLP datasets.

ABSTRACT

As the use of MultiLayer Networks (or MLNs) for modeling and analysis is gaining popularity, it is becoming increasingly important to propose a community definition that encompasses the multiple features represented by MLNs and develop algorithms for efficiently computing communities on MLNs. Currently, communities for MLNs, are based on aggregating the networks into single graphs using different techniques (type independent, projection-based, etc.) and applying single graph community detection algorithms, such as Louvain and Infomap on these graphs. This process results in different types of information loss (semantics and structure). To the best of our knowledge, in this paper we propose, for the first time, a definition of community for heterogeneous MLNs (or HeMLNs) which preserves semantics as well as the structure. Additionally, our basic definition can be extended to appropriately match the analysis objectives as needed. In this paper, we present a structure and semantics preserving community definition for HeMLNs that is compatible with and is an extension of the traditional definition for single graphs. We also present a framework for its efficient computation using a newly proposed decoupling approach. First, we define a k-community for connected k layers of a HeMLN. Then we propose a family of algorithms for its computation using the concept of bipartite graph pairings. Further, for a broader analysis, we introduce several pairing algorithms and weight metrics for composing binary HeMLN communities using participating community characteristics. Essentially, this results in an extensible family of community computations. We provide extensive experimental results for showcasing the efficiency and analysis flexibility of the proposed computation using popular IMDb and DBLP data sets.

Motivation & Objective

• To address the lack of a community definition in heterogeneous multilayer networks (HeMLNs) that preserves both structural and semantic characteristics.
• To overcome information loss in existing methods that aggregate HeMLNs into single-layer graphs before community detection.
• To develop an efficient, extensible framework for computing communities that supports diverse analysis objectives.
• To enable drill-down analysis and flexible composition of communities across multiple network layers.
• To validate the approach on real-world datasets (IMDb and DBLP) with respect to efficiency and analytical versatility.

Proposed method

• Proposes a k-community definition for k connected HeMLN layers that extends traditional single-graph community definitions while preserving semantics and structure.
• Introduces a decoupling-based computation model that divides the community detection process into independent, layered computations to improve efficiency.
• Employs bipartite graph pairings between communities of adjacent layers using two specialized algorithms: MWRM (Maximum Weighted Row Matching) and MWMT (Maximum Weighted Matching on Tripartite graphs).
• Uses customizable weight metrics (ω) and meta-edge computation (ωₑ) to compose communities based on shared characteristics and inter-layer consistency.
• Applies a divide-and-conquer strategy where each layer’s community is computed once (one-time cost), followed by lightweight iterative composition steps (iterative cost).
• Utilizes a modular framework allowing interchangeable pairing algorithms and weight metrics to adapt to different analysis goals.

Experimental results

Research questions

• RQ1 How can a community definition be formally defined for heterogeneous multilayer networks that preserves both structural and semantic integrity?
• RQ2 What computational framework enables efficient detection of multi-layer communities without aggregating layers into a single graph?
• RQ3 How does the decoupled approach compare in performance to monolithic community detection methods in terms of time and scalability?
• RQ4 To what extent can the proposed method support flexible, objective-driven analysis across diverse network layers and entity types?
• RQ5 Can the framework detect meaningful, consistent communities across real-world datasets like IMDb and DBLP, including cross-layer collaborations?

Key findings

• The proposed community definition is compatible with traditional single-graph definitions and extends them to preserve semantics and structure in heterogeneous multilayer networks.
• The decoupled computation model reduces iterative costs to less than one-tenth of the largest one-time cost (e.g., 0.27 sec vs. 5.43 sec), validating high efficiency.
• For the IMDb dataset, the framework detected 3 total consistent cyclic 3-communities and 12 partial matches, revealing meaningful genre-based actor-director groupings such as those involving Leonardo DiCaprio and Danny Boyle.
• The method successfully identified potential unexplored collaborations (e.g., DiCaprio-Swank-Mendes) that have not yet occurred in real-world productions.
• The use of customizable weight metrics (ω) and MWxx algorithms (MWRM, MWMT) enabled flexible, objective-driven composition of communities across layers.
• The framework demonstrated analytical versatility by enabling drill-down analysis and consistent results across multiple layer compositions, even with inconsistent matches highlighting the importance of objective-specific configuration.