Towards a Ground Truth for Affective Classification in Movies

Automatic affective movie classification can be a powerful technology that can facilitate searching, recommending and play listing of movies. In the process of developing a good classification system, the identification of suitable classes and the selection of proper training material is crucial. In this paper we describe the process of developing a reliable ground truth database that will be used for automated classification of emotions conveyed in movies. We identify a list of emotion labels that people commonly use to describe emotions in movies and employ them to annotate movie fragments. The most adequate fragments are then selected to form the final ground truth database which can later be used for affective classification in movies.