Presentation information

General Session

General Session » [GS] J-2 Machine learning

[2H1-J-2] Machine learning: modeling in businesses

Wed. Jun 5, 2019 9:00 AM - 10:20 AM Room H (303+304 Small meeting rooms)

Chair:Makoto Kawano Reviewer:Kohei Miyaguchi

9:00 AM - 9:20 AM

[2H1-J-2-01] Causal Analysis Model of TV Show Attractiveness using Latent Representation Models

〇Yuki Nishimura1, Shimpei Kanazawa1, Tianxiang Yang1, Masayuki Goto1 (1. Waseda University)

Keywords:Deep Learning, TV Show Attractiveness, Dimensionality Reduction, Latent Dirichlet Allocation, Stacked Autoencoder

Over the past 20 years, TV show viewing rates have been declining due to the increase in the types of media and entertainment. Due to this phenomenon, TV broadcasting companies have higher demands to create more attractive TV shows in order to increase TV show viewing rates. To support TV broadcasting companies create more attractive shows, modeling the relationship between TV show viewing rates, celebrities, and content can be strongly effective. In this paper, we model this complex relationship through the utilization of a neural network and latent representation models such as a stacked autoencoder and latent Dirichlet allocation. We demonstrate that measures to increase TV show viewing rates could be devised by conducting experiments with this model.