Joint sentiment aspect model

  • Authors:
  • Noriaki Kawamae

  • Affiliations:
  • Tokyo Denki University, Kanda-Nishiki-cho, Chiyoda-ku, Tokyo, Japan

  • Venue:
  • Proceedings of the 27th Annual ACM Symposium on Applied Computing
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper proposes a generative model that simultaneously detects topics, sentiments, and ratings from review articles. Unlike other sentiment analysis models, our proposal, Joint Sentiment Aspect model (JSA), distinguishes objective and subjective information, for a given item and the corresponding rating, to describe the generative process of each article. For handling these differences, JSA introduces a latent sentiment/aspect class variable into each article and a latent switch variable into each token. These classes allow JSA to project these articles onto a latent space of sentiment/aspect dimensionality. Experiments on review articles show that the proposed model is useful as a generative model.