Multiple Local Community Detection via High-Quality Seed Identification over Both Static and Dynamic Networks

Abstract

Local community detection aims to find the communities that a given seed node belongs to. Most existing works on this problem are based on a very strict assumption that the seed node only belongs to a single community, but in real-world networks, nodes are likely to belong to multiple communities. In this paper, we first introduce a novel algorithm, HqsMLCD, that can detect multiple communities for a given seed node over static networks. HqsMLCD first finds the high-quality seeds which can detect better communities than the given seed node with the help of network representation, then expands the high-quality seeds one-by-one to get multiple communities, probably overlapping. Since dynamic networks also act an important role in practice, we extend the static HqsMLCD to handle dynamic networks and introduce HqsDMLCD. HqsDMLCD mainly integrates dynamic network embedding and dynamic local community detection into the static one. Experimental results on real-world networks demonstrate that our new method HqsMLCD outperforms the state-of-the-art multiple local community detection algorithms. And our dynamic method HqsDMLCD gets comparable results with the static method on real-world networks.

Detect Local Community based on Core Node using Deep Feature Fusion

“…Most existing methods ignore the importance of seed nodes, which will inevitably reduce the quality of the local community obtained under such a premise. The literature [9] reveals the overlapping communities in which the seed node is located by exploring the potential communities of seed nodes; the literature [10] selects a series of high-quality seed nodes with a target by minimizing the conductivity relationship. Although these methods alleviate the problem of "Seed Wasted" to some extent, only considering how to select or alternatively select from a single perspective cannot effectively solve the problem.…”

Local Community Detection in Graph Streams with Anchors

“…Despite the fact that the work in [27] pertains to static networks, we mention the importance of this research while the authors deal with the problem of the maximal a-quasi-clique; from a local community perspective, detecting the communities of a specific node of interest. Furthermore, the authors of [28] proposed the HqsMLCD algorithm in order to detect multiple overlapping communities for a given starting node. The notion of high quality seeds was introduced, which are obtained by the embedded candidate subgraph.…”

Local Community Detection: A Survey

“…Since a starting node can belong to multiple communities, in [97] Liu J. et al extend their static algorithm [16] to the dynamic network to solve this problem. In fact, they are the first to propose a method for multiple LCD in dynamic networks, called HqsDMLCD.…”

“…The random-walk based method proposed by Shi P. et al [88] performs very well compared to other state-of-the-art methods. The authors also provide the source code of the proposed approach and conducted experiments -Adaptation of the static L-metric to dynamic -Zakrzewska A. et al [11] focuses on maintaining the community centered around the seed -Gao W. et al [91] EvoLeaders leader nodes -Nathan E. et al [92] combination of Katz and PageRank centrality, requires a priori knowledge about the size of the community -DiTursi D. et al [93] PHASR temporal conductance measure, weighted -Javadi S. et al [94] leader nodes, degree centrality -Guo K. et al [95] local fitness, node contribution, DyCDNC method -Papadopoulos A. et al [96] -Extension of DiTursi D. et al [93] PHASR algorithm to conform to the Apache Spark engine distributed processing standard -Liu J. et al [97] HqsDMLCD extension of Liu J. et al [16] to dynamic networks, first work for dynamic local multiple communities - [98] LDM-CET Personalized PageRank approach, overlapping -Rigi M. et al [99] method inspired by geometric active contours -Fu D. et al [100] L-MEGA motif-based method, multi-linear PageRank vector by edge filtering -Hu Y. et al [42] D-LM method extension of their static version, not suitable when many nodes are affected by changes -…”

“…Li Y. et al[13],He K. et al[80],Laarhoven T. et al[81],Bian Y. et al[82],Jeub L. et al[83],Bian Y. et al[15],Freitas S. et al[84],He K. et al[85],Zhang Y. et al[86],Fan X. et al[87],Shi P. et al[88],Luo D. et al[89] Snapshot model Takaffoli M. et al[90],Zakrzewska A. et al[11],Gao W. et al[91],Nathan E. et al[92],DiTursi D. et al[93],Javadi S. et al[94],Guo K. et al[95],Papadopoulos A. et al[96],Liu J. et al[97] …”