Identifying Influential Users in Twitter Networks of the Turkish Diaspora in Belgium, the Netherlands, and Germany

Introduction

High-profile political activity and digital campaigns during the ‘Arab Spring’ revolutions and during the Gezi Park protests in Turkey have led scholars focus on the role online social media platforms have played in mass social change and political mobilization.

Social media, with their decentralized network structures, have drastically altered the ways in which information is disseminated and the ways it impacts interpersonal and social communication during political and social events. Mass media are unidirectional (from a media source to a target audience), allowing for the measurement of direct effects (Amaral, Zamora, del Mar, Grandío & Noguera, 2016). However, because of the multidirectional flow of information on the internet, online communication floats in all directions with numerous indirect effects (Amaral et al., 2016). As a result of these differences, the nature of the influence of social media on political action on the internet can be expected to be more complex than it might be within a group of opinion leaders who receive information through mass media and then disseminate their opinion to receptive followers. Through social media, influential people are themselves influenced by other influencers, and this may produce a dense exchange of both information and influence. Ultimately, in the context of internet interactions, opinion leaders both generate and receive influence (Amaral et al., 2016).

These differences have attracted a substantial amount of scholarly interest in a variety of disciplines (Louni & Subbalakshmi, 2014). Of particular interest is identifying trending subjects and principal players on social media platforms, then assessing their influence on social network dynamics through disseminating and broadcasting information online regarding significant political and social events (Louni & Subbalakshmi, 2014)This process of identification and analysis aids in the fast and effective transfer of information (Louni & Subbalakshmi, 2014) and can illuminate how processes such as the dissemination of information and the resultant “cascading behaviors”¹ take place, both of which are important for understanding social events such as political mobilization (Amaral et al., 2016). However, achieving a full understanding of this ‘transfer’ has been challenging and elusive for various reasons: First, many social networks are not entirely observable and in constant flux (Panda, Dehuri & Wang, 2014). For example, people can drop out of a particular discussion or become involved in a new one (Amaral et al., 2016). Second, influencers’ characteristics depend on a range of variables and criteria, including users’ attributes, users’ position in the network structure, users’ number of followers and/or followees, among others (Ma, Li, Bailey, & Wijewickrema, 2017). Third, there is no consensus among the scholarly community on any conclusive method for calculating influence scores. Calculation methods vary depending on the purposes and aims of the individual studies (Li, Zhou, Lü, & Chen, 2014). User influence varies according to changes in interest and behavior over time. Therefore, it is difficult to know how users influence scores should be calculated, with each researcher using his or her own preferred method (Li et al., 2014). These differences in calculation methods makes it difficult to compare data from study to study or undertake a meta-analysis. Studies measuring user influence have focused on user attributes, network structure, user importance, user interactions, and network position of user (Ma et al., 2017).