Influence Of The User Importance Measure On The Group Evolution Discovery

TL;DR

This paper examines how different user importance measures affect the accuracy of social group evolution detection, finding that global measures like PageRank improve results over local measures.

Contribution

It introduces an analysis of the impact of various user importance measures on the group evolution discovery method, highlighting the effectiveness of global measures.

Findings

Global importance measures yield more precise group evolution results.

PageRank outperforms local measures like degree centrality.

Using no importance measure results in less accurate detection.

Abstract

One of the most interesting topics in social network science are social groups. Their extraction, dynamics and evolution. One year ago the method for group evolution discovery (GED) was introduced. The GED method during extraction process takes into account both the group members quality and quantity. The quality is reflected by user importance measure. In this paper the influence of different user importance measures on the results of the GED method is examined and presented. The results indicate that using global measures like social position (page rank) allows to achieve more precise results than using local measures like degree centrality or no measure at all.