ESC: Edge-attributed Skyline Community Search in Large-scale Bipartite Graphs

TL;DR

This work defines Edge-attributed Skyline Communities (ESC) in large-scale bipartite graphs by uniting ($\alpha$, $\beta$)-core cohesiveness with multi-dimensional edge attributes through a skyline framework. It introduces peeling and expanding algorithms that respectively prune the search space and grow candidate ESCs, with specialized handling across dimensionalities $d=1,2,3$ and a recursive strategy for $d>3$, all while ensuring non-dominance of results. The approach is validated on six real datasets, demonstrating improved efficiency, scalability, and the ability to produce diverse, high-quality communities containing query vertices, outperforming single-dimension or purely structural models in multi-attribute settings. The ESC model thus enables precise, multi-attribute community discovery with broad applicability to recommendation, transportation, and bioinformatics scenarios.

Abstract

Due to the ability of modeling relationships between two different types of entities, bipartite graphs are naturally employed in many real-world applications. Community Search in bipartite graphs is a fundamental problem and has gained much attention. However, existing studies focus on measuring the structural cohesiveness between two sets of vertices, while either completely ignoring the edge attributes or only considering one-dimensional importance in forming communities. In this paper, we introduce a novel community model, named edge-attributed skyline community (ESC), which not only preserves the structural cohesiveness but unravels the inherent dominance brought about by multi-dimensional attributes on the edges of bipartite graphs. To search the ESCs, we develop an elegant peeling algorithm by iteratively deleting edges with the minimum attribute in each dimension. In addition, we also devise a more efficient expanding algorithm to further reduce the search space and speed up the filtering of unpromising vertices, where a upper bound is proposed and proven. Extensive experiments on real-world large-scale datasets demonstrate the efficiency, effectiveness, and scalability of the proposed ESC search algorithms. A case study was conducted to compare with existing community models, substantiating that our approach facilitates the precision and diversity of results.

Figures (10)

• Figure 1: Example of communities containing the query.
• Figure 2: An example of peeling in case $d=2$.
• Figure 3: An example of expanding in case $d=2$.
• Figure 4: Recursive decomposition approach.
• Figure 5: An example of expanding in case $d=3$.

Theorems & Definitions (14)

• Definition 1
• Definition 2
• Definition 3
• Definition 4
• Definition 5
• Example 1
• Lemma 1
• Example 2
• Lemma 2
• Lemma 3