Deepdense: Enabling Node Embedding to Dense Subgraph Mining

21 Pages Posted: 29 Jan 2023

See all articles by Walid Megherbi

Walid Megherbi

affiliation not provided to SSRN

Mohammed Haddad

affiliation not provided to SSRN

Hamida Seba

University of Claude Bernard Lyon 1

Abstract

Dense subgraphs convey important information and insights about a graph structure. This explains why dense subgraph mining is a  problem of key interest that arises in several tasks and applications such as graph visualization, graph summarization, graph clustering, and complex network analysis. It is a hard problem that has been intensively addressed by the data mining community.\\In this paper, we propose a deep learning approach that enumerates all occurrences of dense subgraphs in a graph without any constraints or limitations on their size. More precisely, we enrich exiting structural node embedding with extra information, computed on node neighborhoods, to effectively capture their belonging to  dense subgraphs. We evaluate our approach on several datasets to attest its efficiency on two main applications: graph summarization and graph clustering.

Keywords: Dense subgraph mining, node embedding, graph learning

Suggested Citation

Megherbi, Walid and Haddad, Mohammed and Seba, Hamida, Deepdense: Enabling Node Embedding to Dense Subgraph Mining. Available at SSRN: https://ssrn.com/abstract=4341593 or http://dx.doi.org/10.2139/ssrn.4341593

Walid Megherbi

affiliation not provided to SSRN ( email )

No Address Available

Mohammed Haddad

affiliation not provided to SSRN ( email )

No Address Available

Hamida Seba (Contact Author)

University of Claude Bernard Lyon 1 ( email )

43 Bl du 11 novembre 1918
Lyon, Villeurbanne cedex 69622
France

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
43
Abstract Views
197
PlumX Metrics