Designing a Comic Exploration System Using a Hierarchical Topic Classification of Reviews
Abstract
The purpose of our research is information access support based on the contents of a comic book. For this purpose, it is necessary to obtain information related to the story and the characters. In our previous research, we extracted the information using a review sentence and built a comic search system based on the extraction results. This system determined the relationship between comics using the TF-IDF method. However, the TF-IDF method cannot take into account the meaning of each word included in the review sentence. Therefore, this system may not be able to provide accurate results based on comic relationships. In this study, we analyze the review sentence using a hierarchical topic classification. On the other hand, the extracted topics often contain words that hinder the user from guessing the contents of the comic. Especially, the named entities defined in the comic tend to cause the problem. The amount of information obtained by the user varies greatly based on the user’s knowledge about the named entities. Therefore, we investigated the influence of the named entity extracted from the comic reviews as the feature words. Our experiment revealed that the user’s understandability was improved when the proportion of the named entities was decreased. Furthermore, we build an exploratory search system based on the topic.
References
David. M. Blei, Thomas. L. Griffiths, and Michael. I. Jordan. The nested chinese restaurant process and bayesian nonparametric inference of topic hierarchies. Journal of the ACM, Vol. 57, No. 2, 2010, Article 7.
David. M. Blei, Andrew. Y. Ng, and Michael. I. Jordan. Latent dirichlet allocation. Journal of Machine Learning Research, Vol. 3, 2003, pp. 993–1022.
Sheena. S. Iyenger and Mark. R. Lepper. When choice is demotivating: Can one desire too much of a good thing? Journal of Presonality and Social Psychology, Vol. 79, No. 6, 2000, pp. 995–1996.
White. W. Ryen and Roth A. Resa. Exploratory search: Beyond the query-response paradigm. Morgan & Claypool Publishers, 2009.
Robert. S. Taylor. Question-negotiation and information seeking in libraries. College & Research Libraries, Vol. 29, No. 3, 1968, pp. 178–194.
Yamashita, R. and Matsushita, M. Content discrimination of comics based on users’ reviews. The Third Asian Conference ofn Information Systems, 2014, pp.79–85.
Okada, S. and Arakawa, T. A proposal on support system for reading of novels using question answering technology. Japan Society for Fuzzy Theory and Intelligent Informatics, Vol. 27, No. 2, 2015, pp. 608–615, in Japanese.
Iwama, Y. and Mihara, T. and Nagamori, M. and Sugimoto, S. Facet-based visualization of a Manga collection based on ontology and LOD resources. IEICE Human Communication Group Symposium, 2014, pp. 357–361, in Japanese.
Yamashita, R. and Park, B. and Matsushita, M. Supporting exploratory information access based on comic content information. Transaction of the Japanese Society for Artifical Intelligence, Vol. 32, No. 1, 2017, pp. WII–D 1–11.
Lafferty, J., McCallum, A. and Pereira, F. C. Conditional random fields: Probabilistic models for segmenting and labeling sequence data Proceedings of the 18th International Conference on Machine Learning, 2001, pp. 282–289.
Sekine, S., Sudo, K. and Nobata, C. Extended named entity hierarchy Proceedings ofthe 3rd International Conference on Language Re-sources and Evaluation, 2002, pp. 1818–1824.
Kiyotaka, U., Qing, M., Masaki, M., Hiromi, O. and Hitoshi, I. Named Entity extraction based on a maximum entropy model and transformation rules. Proceedings of the 38th Annual Meeting on Association for Computational Linguistics, pp. 326–335, 2000.
