Identifying Opinion Drivers on Social Media

Social media is increasingly playing a central role in commercial and political strategies, making it an imperative to understand its dynamics. In our work, we propose a model of social media as a “marketplace of opinions.” Online social media is a participatory medium where several vested interests invest their opinions on disparate issues, and actively seek to establish a narrative that yields them positive returns from the population. This paper focuses on the problem of identifying such potential “drivers” of opinions for a given topic on social media. The intention to drive opinions are characterized by the following observable parameters: (a) significant level of proactive interest in the issue, and (b) narrow focus in terms of their distribution of topics. We test this hypothesis by building a computational model over Twitter data. Since we are trying to detect an intentional entity (intention to drive opinions), we resort to human judgment as the benchmark, against which we compare the algorithm. Opinion drivers are also shown to reflect the topical distribution of the trend better than users with high activity or impact. Identifying opinion drivers helps us reduce a trending topic to its “signature” comprising of the set of its opinion-drivers and the opinions driven by them.