Predictive Analytics on Emotional Data Mined from Digital Social Networks with a Focus on Financial Markets

This dissertation is a cumulative dissertation and is comprised of five articles. User-Generated Content (UGC) comprises a substantial part of communication via social media. In this dissertation, UGC that carries and facilitates the exchange of emotions is referred to as “emotional data.” People “produce” emotional data, that is, they express their emotions via tweets, forum posts, blogs, and so on, or they “consume” it by being influenced by expressed sentiments, feelings, opinions, and the like. Decisions often depend on shared emotions and data – which again lead to new data because decisions may change behaviors or results. “Emotional Data Intelligence” ultimately seeks an answer to the question of how all the different emotions expressed in public online sources influence decision-making processes. The overarching research topic of this dissertation follows the question whether network structures and emotional sentiment data extracted from digital social networks contain predictive information or they are just noise. Underlying data was collected from different social media sources, such as Twitter, blogs, message boards, or online news and social networking sites, such as Xing. By means of methodologies of social network analysis (SNA), sentiment analysis, and predictive analysis the individual contributions of this dissertation study whether sentiment data from social media or online social networking structures can predict real-world behaviors. The focus lies on the analysis of emotional data and network structures and its predictive power for financial markets. With the formal construction of the data analyses methodologies introduced in the individual contributions this dissertation contributes to the theories of social network analysis, sentiment analysis, and predictive analytics.