Sentiment classification and computing for online reviews by a hybrid SVM and LSA based approach

For the current online reviews sentiment classification method, there are some problems such as serious text sparseness and coarse granularity of sentiment calculation. In this paper, the emotion in online reviews is divided into four categories: happiness, hope, disgust, and anxiety. Based on the combination of cognitive evaluation theory and sentiment analysis, a novel approach that combines a well-known techniques to sentiment classification, ie, support vector machine and the latent semantic analysis, was proposed. Based on the approach, this paper explored the influence of these four kinds of emotions on the helpfulness of online reviews, examined the moderating effects of emotion on the helpfulness of online reviews under the two types of products. The experimental results showed that this model could effectively conduct multi-emotion fine-grained computing for online reviews, improve the accuracy and computational efficiency of sentiment classification. The final empirical analysis found that happiness and disgust emotion had significant positive impact on the helpfulness of online reviews, while on the other hand anxiety emotion had significant negative influence. The algorithm and its empirical conclusions provide useful theoretical basis and reference for the company to optimize marketing strategy and improve customer relationship under web 2.0.