Resumen

MUNIZ-CUZA, Carlos E.  y  ORTEGA-BUENO, Reynier. Aspect-Based Polarity Classification Method on Products Re-view. Rev cuba cienc informat [online]. 2016, vol.10, n.1, pp.141-151. ISSN 2227-1899.

ABSTRACT This paper presents a method for aspect-based sentiment analysis on user products reviews. The most outs-tanding feature in this proposal is the automatic building of domain-depended sentiment resource using Latent Semantic Analysis. The proposed method can be adapted to different textual units such as bigrams and trigrams and is language independent. The aspect term polarity classification is carried out in two phases: opinion words and phrases extraction and polarity classification. The extraction phase involve the search of surface and parse feature of the aspect, getting polarities scores of the textual units generated on the previous phase. Finally, the polarity of the aspect, in a given review, is determined from the positive and negative scores of each words and phrases extracted. The results obtained by the approach are encouraging if we consider that the construction of the domain-dependent polarity lexicon is performed fully automatic.

Palabras clave : polarity; latent semantic analysis; surface feature; parse feature.

        · resumen en Español     · texto en Español     · Español ( pdf )