CASUS Institute Seminar, Zongru Shao, CASUS Görlitz
Emotion is a crucial indicator of the mental state influenced by environmental cognition, health, and intention. Modeling emotional patterns from daily communications is significant for behavioral studies. This talk presents an analytical system that combines psycho-linguistics, statistical modeling, and computational algorithms to extract, analyze, and rank users’
emotions from the text. The system includes a twelve-emotion categorization to identify a balanced set of six positive emotions and their negative counterparts. It uses a lexicon-based algorithm to extract emotions by combining thesaurus and parts-of-speech. Then, it utilizes a Markov chain model to analyze the time-dependent emotion stochastic process from four
aspects: repetition, revisits, steady states, and emotional stability.
Finally, the system ranks the participants from the most emotional to the least. A case study is conducted on the Enron email dataset. Emotions are modeled for 22 executives who were deemed guilty in the court of law in comparison with 21 who were non-guilty. The modeling provides a detailed and distinct pattern between the guilty and non-guilty, compared to the four-emotion LIWC (Linguistic Inquiry and Word Count) categorization and the six Ekman’s emotions. The system may also provide insights for other related research such as depression and mental health.