Analyzing Emotional Statements – Roles of General and Physiological Variables

The present task collects different statistics of emotions based on the combinations of general variables (intensity, timing and longevity) and physiological variables (psycho-physiological arousals) from the situational statements of the ISEAR (International Survey on Emotion Antecedents and Reactions) dataset. The individual as well as combinational roles of different variables are analyzed. Some interesting observations and insights are found with respect to emotions. The statements of similar emotions are clustered according to different combinations of the variables. Each of the statements of a cluster is passed through two types of emotion tagging systems, a lexicon based baseline system followed by a supervised system. Due to the difficulty of incorporating knowledge regarding physiological variables, the supervised system only considers the roles of general variables from textual statements. The roles of the general variables are played by intensifiers, modifiers and explicitly specified temporal and causal discourse markers. The evaluation indicates that the supervised system based on general variables produces satisfactory results in identifying emotions.