A Survey on Feature-Sentiment Classification Techniques

Abstract

As internet growing exponentially, the online purchase is proportionally increasing its all around the world. The e-commerce and product selling websites are providing a rich variety of product to be sold. As the quality of product has much impact on its sell, the e-commerce websites tends to take public opinion on the product in terms of consumers feedback, we call it as reviews. These reviews provide much knowledge about the product as the consumers are motivated to write their reviews about the product, more precisely saying, consumer writes their opinion about product’s specifications or product’s features. These public opinions can then be analyzed by the consumers and vendor to make the required manufacturing changes to the product to increase its quality. The Feature Mining along with Sentiment Analysis techniques can be applied to achieve product’s feature and public opinion on these features. Here in this paper we are motivated by the scenario as mentioned above. We had a survey on the different techniques that can be used to mine products feature and classifying those feature along with the sentiment classification on the determined features. The public sentiments can be classified as negative, positive and neutral sentiments. Data Mining provides a rich set of Machine Learning Algorithms which in turn can be used as Sentiment Classifier. After analyzing feature-sentiment techniques, we then studied the feature classification by using its overall sentiment and influence on the product sell.