Triangles as basis to detect communities: an application to Twitter's network

TL;DR

This paper introduces a novel community detection algorithm based on triangle structures in social networks, complemented by an interactive visualization tool, and evaluates its effectiveness on both benchmark and real Twitter data.

Contribution

The paper presents a new triangle-based community detection algorithm and an interactive visualization approach, applied to Twitter data and benchmark networks.

Findings

Effective community detection on benchmark data

Successful application to Twitter network data

Enhanced visualization of social network communities

Abstract

Nowadays, the interest given by the scientific community to the investigation of the data generated by social networks is increasing as much as the exponential increasing of social network data. The data structure complexity is one among the snags, which slowdown their understanding. On the other hand, community detection in social networks helps the analyzers to reveal the structure and the underlying semantic within communities. In this paper we propose an interactive visualization approach relying on our application NLCOMS, which uses synchronous and related views for graph and community visualization. Additionally, we present our algorithm for community detection in networks. A computation study is conducted on instances generated with the LFR [9]-[10] benchmark. Finally, in order to assess our approach on real-world data, we consider the data of the ANR-Info-RSN project. The latter addresses community detection in Twitter.