Emotion argumentation

Abstract

Argumentation, constituting of major component of human intelligence is considered as a process where the arguments are constructed as well as tackled. Argumentation is a collection of propositions called “Premises” except one which is termed as “Conclusion”. If we identify argumentation from the perspectives of emotions, it means to examine whether consistency is conveyed from a set of premises to its corresponding conclusion or not. In the present task, we have developed a rule based baseline system followed by a machine learning frame work. Two types of different corpora, ECHR (European Court of Human Rights) and the Araucaria Database were used for experiments. We used the Bayes' theorem to find the effects of various emotions in identifying conclusion from the set of given premises with the help of argumentation. We have employed the Naïve Bayes, Sequential Minimal Optimization (SMO) and Decision Tree classifiers in our machine learning frame work and evaluated the results of the rule based system by manual experts. The evaluation achieves the maximum F-Score of 0.874 and 0.649 for premises and conclusion in case of rule based system whereas 0.958 and 0.815 for the Naïve Bayes, 0.893 and 0.458 for the SMO and 0.951 and 0.957 for the Decision Tree classifiers, respectively.