Anime clustering for automatic classification and configuration of demographics

Abstract

The cultural industry assumed greater relevance as a productive system and expanded its market share with different forms of reception, transmission, and communication with the public, increasingly using the so called classification and recommendation algorithms and manipulation of mass processed data, which do not require cyber-physical systems for cataloging andconstant feedback from all parties involved for cataloging. In this regard, this paper proposes a methodology to support the classification and creation of corresponding groups, automatically, of cultural productions of certain segments through Robot Process Automation (RPA) techniques, to first extract public data created by fans of certain cultural segments, and Latent Dirichlet Allocation (LDA), for the clustering of these productions based on the data of the terms extracted by RPA. As a case study for this proposal, we specifically observed the anime market, defined as an originally Japanese cultural product with high fan engagement and high annual production scale, supported by data obtained from two public databases data: MyAnimeList and AniDB, built collaboratively by fans. The application of the methodology allowed the automatic classification of anime, grouping them into topics that allow the proposal of a new demography of products of this genre in relation to the current one, providing a greater level of detail and allowing to contemplate the expansion of new themes.