Analysis of Four-scene Comics Story Dataset based on natural language processing

Abstract

Comic computing is a branch of computing dealing with comics in engineering. Although comics are multi-modal data with natural languages and pictures, numerous studies in the field focus on images in comics rather than the comic content. To make models understand the contents in comics, we should deal with natural languages in comics in the form of character words. We used a dataset which was suitable for analyzing comic contents and previously proposed two tasks: sentiment analysis and variety analysis. However, two tasks did not go well because of the number of data. We demonstrated using data augmentation and analyzed the results to determine the feasibility of computers understanding comics.