Review Mining - SDM 2011        


Exploiting Coherence in Reviews for Discovering Latent Facets and Associated Sentiments

Himabindu Lakkaraju, Chiranjib Bhattacharyya, Indrajit Bhattacharya, Srujana Merugu

 

Best Paper Award at SDM 2011

Paper Download

Presentation Download

Abstract

  • With online expression of sentiment becoming freely available in large volumes, and customers increasingly relying on reviews to decide on products, the demand for sentiment analysis techniques has been increasing tremendously. While early sentiment analysis techniques focused on determining an overall sentiment score for each review, more recent approaches try to discover reasons for satisfaction or dissatisfaction and associate sentiments with specific product facets. This kind of facet-based sentiment analysis of customer reviews involves discovering the latent facets, sentiments and their associations. Traditional facet-based sentiment analysis algorithms typically perform the various tasks in sequence, and fail to take advantage of the mutual reinforcement of the tasks. Additionally, inferring sentiment levels typically requires domain knowledge or human intervention. In this talk, we present a series of probabilistic models that jointly discover latent facets and sentiment topics, and also order the sentiment topics with respect to a multi-point scale, in a language and domain independent manner. This is achieved by simultaneously capturing both short-range syntactic structure and long range semantic dependencies between the sentiment and facet words. The models further incorporate coherence in reviews, where reviewers dwell on one facet or sentiment level before moving on, for more accurate facet and sentiment discovery. For reviews which are supplemented with ratings, our models automatically order the latent sentiment topics, without requiring seed words or domain knowledge. For reviews which are supplemented with ratings, our models automatically order the latent sentiment topics, without requiring seed-words or domain-knowledge.

Dataset

  • The complete set of reviews from amazon that we used can be downloaded here.

  • The labelled reviews can be downloaded here.

Contact

  • We would be happy to answer your queries. Please mail your comments/queries to the email address : klakkara@in.ibm.com (or) lvhimabindu@gmail.com

References:

  1. Himabindu Lakkaraju, Chiranjib Bhattacharyya, Indrajit Bhattacharya, Srujana Merugu - Exploiting Coherence in Reviews for Discovering Latent Facets and Associated Sentiments , SIAM International Conference on Data Mining, 2011, Phoenix, Arizona