Semantic approaches to fine and coarse-grained feature-based opinion mining

  • Authors:

  • Alexandra Balahur;Andrés Montoyo

  • Affiliations:

  • Department of Software and Computing Systems, University of Alicante, Alicante, Spain;Department of Software and Computing Systems, University of Alicante, Alicante, Spain

  • Venue:
  • NLDB'09 Proceedings of the 14th international conference on Applications of Natural Language to Information Systems
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

Feature-based opinion mining from product reviews is a difficult task, both due to the high semantic variability of opinion expression, as well as because of the diversity of characteristics and sub-characteristics describing the products and the multitude of opinion words used to depict them. Further on, this task supposes not only the discovery of directly expressed opinions, but also the extraction of phrases that indirectly or implicitly value objects and their characteristics, by means of emotions or attitudes. Last, but not least, evaluation of results is difficult, because there is no standard corpus available that is annotated at such a fine-grained level and no annotation scheme defined for this purpose. This article presents our contributions to this task, given by the definition and application of an opinion annotation scheme, the testing of different methodologies to detect phrases related to different characteristics and the employment of Textual Entailment recognition for opinion mining. Finally, we test our approaches both on the built corpus, as well as on an ad-hoc built collection of reviews that we evaluate on the basis of the stars given. We prove that our approaches are appropriate and give high precision results.