Category:

Guide

Duration: 1:06:32

84 views

2

As the surge of communication affordances over social media continues unabated, it has greatly expanded our horizons of putting the generated interactional information to good use. Almost inconceivable scarcely a decade ago, on one hand, it has enabled researchers to study social processes at extremely large-scales, while on the other it has streamlined the end user experience in terms of the ability to explore information on timely happenings essentially anytime, anywhere. However, with several terrabytes of such information generated every day, we are presented with the daunting question: how do we identify those pieces of information that really matter? In this talk, we will investigate relevance of entities in the social media context from two perspectives: (1) the context of studying macroscopic social processes, and (2) the context of enhancing our online social experience. First, we will discuss the problem of inferring relevant network structures from observed communication events (e.g. emails). Based on empirical studies, we will investigate how to identify meaningful social ties that can enable the understanding of larger sociological constructs, such as homophily or community detection. In the second part, digressing from the state-of-the-art, we will make an excursion into devising information sampling techniques that benefit from constructs in the cognitive sciences, dealing with measures of human information processing. Based on an elaborate user study spanning over more than a billion pieces of Twitter content, we will discuss how our methods of identifying relevant social media information can inform the design of better interfaces and socio-technical systems in general.