A Discriminative Approach to Sentiment Classification

摘要

Due to the explosive growth of user-generated contents, understanding opinions (such as reviews on products) generated by Internet users is important for optimizing business decision. To achieve such understanding, this paper investigates a discriminative approach to classifying opinions according to sentiments. The discriminative approach builds a model with the prior knowledge of the categorization information in order to extract meaningful features from the unstructured texts. The prior knowledge includes ratio factors to reinforce terms’ sentiment polarity by using TF-IDF, short for term frequency-inverse document frequency. Experimental results with four datasets show the proposed approach is very competitive, compared with some of the previous works.