查看文章

In this work, the problem of classifying documents not by topic, but by overall sentiment, eg, determining whether a review is positive or negative is investigated based on Gaussian Mixture Model (GMM) and Feed Forward Neural Network (FFNN). Using movie reviews as data, it is found in this work that the summarization step using FFNN model for the movie reviews increases the sentiment classification accuracy. The proposed approach is a trainable classifier which uses TFIDF feature to train Gaussian Mixture Model (GMM) in order to construct a sentiment classifier model. An automatic text summarizer is exploited before the use of GMM to increase the sentiment classification accuracy.