Discerning Emotions of Bloggers based on Topics–a Supervised Coreference Approach in Bengali

Discerning Emotions of Bloggers based on Topics–a Supervised Coreference Approach in Bengali

This paper presents an approach to identify the users'emotions on different topics provided in Bengali blog posts. The identification of emotion holder, emotion topic along with emotional expression aims to develop the baseline system. The input vectors containing emotional expressions and topics with respect to the corresponding emotion holders are prepared from the annotated blog posts. The feature vectors consist of different syntactic, semantic, rhetoric and overlapping features are employed in a supervised system to identify the co reference of the emotion holder with the corresponding emotional expression and topic. Evaluation metric (Krippendorff'sα) achieves the agreement scores of 0.53 and 0.67 for the baseline and supervised coreference classification systems respectively. The emotions for different topics with respect to each of the blog users represent the topic dependent users'emotional views.