Development of review selection system reflecting user preference using svm

Abstract

In recent years, along with the spread of e-commerce sites, reviews of goods and services are often posted on the web. In a “review”, a user comments and evaluates a specific product or service. Examples of e-commerce sites are Amazon(https://www.amazon.co.jp/ ) and Netflix(https://www.netflix.com/jp/ ). Users make selections based on the performance and review of goods and services. Especially in the case of items which cannot be compared by performance, selection is difficult. Therefore, the review is an important selection factor of the item for a user. However, the perspectives of reviews vary depending on the different preferences of people. For that reason, a review that refers to the content a user wants to know is difficult to find quickly. For this study, we developed a system that distinguishes review sentences based on a user preference using machine learning. Then the system presents them to the user. This information will support decision making by users of e-commerce sites and will contribute to the improvement of information collection efficiency.