On the discovery of social roles in large scale social systems

Abstract

The social role of a participant in a social system is a label conceptualizing the circumstances under which she interacts within it. They may be used as a theoretical tool that explains why and how users participate in an online social system. Social role analysis also serves practical purposes, such as reducing the structure of complex systems to relationships among roles rather than alters, and enabling a comparison of social systems that emerge in similar contexts. This article presents a data-driven approach for the discovery of social roles in large scale social systems. Motivated by an analysis of the present art, the method discovers roles by the conditional triad censuses of user ego-networks, which is a promising tool because they capture the degree to which basic social forces push upon a user to interact with others. Clusters of censuses, inferred from samples of large scale network carefully chosen to preserve local structural properties, define the social roles. The promise of the method is demonstrated by discussing and discovering the roles that emerge in both Facebook and Wikipedia. The article concludes with a discussion of the challenges and future opportunities in the discovery of social roles in large social systems.

Regular Equivalence for Social Networks

“…These approaches are in general much more mathematically oriented than ours, focusing on strict and rigorous scientific analysis. They have been generalized in many directions, e.g., fuzzy graphs [28], applying data-driven [29], and unsupervised role learning methods [30], some of these building upon massive amounts of data in contrast to our approach. New methods extend the ideas of equivalence through the use of additional measures when defining roles, like geodesic distance, modularity, community structure, and cliques [31][32][33], which make use of complimentary concepts beyond our definitions.…”

Generating global network structures by triad types

“…Understanding such mechanisms might also contribute to the development of statistical tests for assessing the significance of the blockmodel structure obtained by simulations [9] or to predict missing links in partly known networks [10]. In this context, the triad census has already been used to detect the social rules of a participant in a social system [11],…”

Towards a Modeling Framework of Social Contexts, Roles and Relations for Acquiring Role-Specific Rules

“…[18] introduces a data mining model for social role discovery and attribution in Internet forums. [19] presents a data-driven method to discover social roles that exist in largescale online social systems. [20] builds a framework for detecting user occupations on microblogging platforms.…”