Quantitative analysis of evolution and role behaviour in an online social network

Abstract

A number of online discussion groups have a long history where individual users are found to participate over long time ranges. These groups therefore offer the possibility to test hypothesis such as preferential attachment on such time scales. The focus of this thesis is in particular to develop quantitative indicators for the type of discussion (e.g. philosophical or technical) and the self-defined roles of the participants [Chang 2002].

Investigations into these two groups confirm similarities and differences in statistical properties of the networks. The degree distribution, network size, clustering and betweeness are all examined. New measures introduced, include the reply count and positions of the posts and globally each group is compared to each other.

Top actors of both groups are selected exploring their individual networks, through the use of Gephi an open source graphical manipulation software, [Bastian 2009].

Through analysing the discussions three roles are observed, the answer role, question role and discussion role. Developing indicators for these roles observe quantitatively how these roles are classified.

Date of Award	2013
Original language	English
Awarding Institution	Coventry University
Supervisor	Christian von Ferber (Supervisor)