Abstract
The aim of this research is to develop a search system for comics based on the personalities of appearing characters. For this purpose, this paper describes the classification of characters using egograms, which are used to classify personalities. In the proposed method, texts that express a comic book character's personality are acquired from web resources, and semantic vectors are allocated based on these texts using egograms. The resulting egogram pattern is used to estimate typical properties. Our experiment reveals that the performance accuracy of this classification method is 55.0%.
Information
Book title
Proc. the 4th International Symposium on Affective Science and Engineering, and the 29th Modern Artificial Intelligence and Cognitive Science Conference
Date of issue
2018/05/31
Date of presentation
2018/06/01
Location
Spokane, WA, USA (Eastern Washington University, Spokane campus)
Citation
Byeongseon Park, Kanae Ibayashi, Mitsunori Matsushita. Classifying Personalities of Comic Characters Based on Egograms, Proc. the 4th International Symposium on Affective Science and Engineering, and the 29th Modern Artificial Intelligence and Cognitive Science Conference, No.B3-2, 2018.