Deep Learning Applied on Refined Opinion Review Datasets

Abstract

Deep Learning has been successfully applied in hard to solve areas, such as image recognition and audio classification. However, Deep Learning has not yet reached the same performance when employed in textual data, including Opinion Mining. In models that implement a deep architecture, Deep Learning is characterized by the automatic feature selection step. The impact of previous data refinement in the pre-processing step before the application of Deep Learning is investigated to identify opinion polarity. This refinement includes the use of a classical procedure of textual content and a popular feature selection technique. The results of the experiments overcome the results of the current literature with the Deep Belief Network application in opinion classification. In addition to overcoming the results, their presentation is broader than the related works, considering the change of parameter variables. We prove that combining feature selection with a basic preprocessing step, aiming to increase data quality, might achieve promising results with Deep Belief Network implementation.