CNN seminar - Dr Elizabeth Leicht

Abstract:

Technology now mediates a significant fraction of social interactions.  Computational social science is using the large data-sets of user interactions drawn from such communication channels as SMS texts, mobile phone voice calls, and online social networks to provide important insights into the structure and dynamics of social networks involving thousands or even millions of users. These communication or behavioural based data-sets document numerous instances of human interaction, but information on the real attributes of, or the actual relationships linking, interacting individuals is frequently lacking.  Attribute data has the potential to refine our understanding of social interactions that are difficult to qualify based on behavioural data alone.   Using a unique 18-month mobile phone data-set in combination with supporting self-reported attribute data from the mobile phone users our work unifies behavioural data studies with more traditional social network analysis based on survey data.   Our analysis reveals the correspondence between social interaction attribute data reported by study participants (ties), including the participants' valuation of tie strength, with the behavioural interactions occurring via mobile phones voice calls and SMS text messages.  We show that variations in perceived social tie strength translate into discernible differences in communication behaviour via mobile phone voice calls and SMS text messages.  We also connect disparities in behavioural data across the two communication channels to self-reported tie attributes.