Representation and Analysis of Twitter Activity: A Dynamic Network Perspective

Abstract

Online interaction networks are highly dynamic. They provide opportunities to share information more widely and faster than ever before, but result in rather complicated trajectories of information flow that present new challenges for modeling and analysis. Collecting the right data, processing it appropriately and determining which networks to analyze are some of the challenges we face. The ease with which data can be sourced through Twitter's API has resulted in a disproportionate number of studies analyzing 'big data'. This focus has led researchers to overlook the importance of 'small data': traditional methods of data collection, such as survey and experimental studies. Clever data extraction and processing capabilities are absolutely necessary to deal with the enormous quantity of data generated by Internet-mediated interactions. However, the richness at the level of the individual may be overlooked if big data approaches are exclusively used, potentially resulting in inappropriate generalizations and conclusions. In this paper we study results from a qualitative study on Twitter users' behavior and combine them with dynamic network analysis to investigate the manifestations of Twitter relations.