Generative Models for Sentiment Analysis and Opinion Mining

  • Wang H
  • Zhai C
N/ACitations
Citations of this article
22Readers
Mendeley users who have this article in their library.
Get full text

Abstract

This chapter provides a survey of recentwork on using generative models for sentiment analysis and opinion mining. Generative models attempt to model the joint distribution of all the relevant data with parameters that can be interpreted as reflecting latent structures or properties in the data. As a result of fitting such a model to the observed data, we can obtain an estimate of these parameters, thus “revealing” the latent structures or properties of the data to be analyzed. Such models have already been widely used for analyzing latent topics in text data. Some of the models have been extended to model both topics and sentiment of a topic, thus enabling sentiment analysis at the topic level.Moreover, newgenerative models have also been developed to model both opinionated text data and their companion numerical sentiment ratings, enabling deeper analysis of sentiment and opinions to not only obtain subtopic-level sentiment but also latent relative weights on different subtopics. These generative models are general and robust and require no or little human effort in model estimation. Thus they can be applied broadly to perform sentiment analysis and opinion mining on any text data in any natural language.

Cite

CITATION STYLE

APA

Wang, H., & Zhai, C. (2017). Generative Models for Sentiment Analysis and Opinion Mining (pp. 107–134). https://doi.org/10.1007/978-3-319-55394-8_6

Register to see more suggestions

Mendeley helps you to discover research relevant for your work.

Already have an account?

Save time finding and organizing research with Mendeley

Sign up for free