E-Commerce Comment Sentiment Classification Based on Deep Learning

Abstract

With the development of e-commerce, comments on e-commerce platform has emerged as an important source of information for understanding customers' attitude. Sentiment classification is a direction of NLP (Natural Language Processing), which focuses on classify the comments into positive class, neutral class, negative class according to the polarity of sentiment. Deep learning-based method for sentiment classification has been a mainstream due to its outstanding performance. Sentiment classification of e-commerce comments can help merchants and e-commerce platforms understand customer preferences and needs to improve service quality and customer's satisfaction. In this paper, in order to annotate the comment dataset efficiently, we propose a method for annotation based on artificial and emotional lexicon. This paper consists of two parts of works. First of all, we adopt the artificial plus emotional lexicon method to annotate the original comment dataset, which can remarkably improve the efficiency and accuracy of the annotation. And then we enter the labeled comment set into the deep learning model and the SVM model. The results show the accuracy of sentiment classification model based on deep learning is significantly higher than that of SVM model.