Classification and Summarization of Pros and Cons for Customer Reviews

As e-commerce is becoming more and more popular, the number of customer reviews for online products grows rapidly. For a popular product, there can be hundreds of reviews. This makes it difficult for a potential customer to read all of them in order to get as much information as possible and to make a decision on purchasing. Therefore, a summarization of product reviews would make purchase more convenient and reliable. The conventional way of summarizing a review is to select or rewrite a subset of the original sentences from the review, which is inefficient. In this paper, we propose to summarize all customers’ reviews of a product as a list of phrases named pros and cons list, which can be perceived and understood at a glance. We employ a score algorithm which considers the strength of a word towards positive or negative orientation to calculate and weigh the sentiment of a sentence. To assess our algorithm, a number of existing classifiers are also presented. Our experimental results show that our Sentence Weight classifier is more accurate and effective than those compared.