Định hướng mới trong trong hệ thống gợi ý

Abstract

Most of the current suggestion system studies focus on collaborative filtering algorithms, but not on the content-based filtering algorithms solution. While the recommender systems suggested by the collaborative filtering algorithm team only focused on evaluating and giving the most general result of a particular item. But in reality, users are more interested in the details of each feature of the item. In this study, we place an interest in automatically analyzing customer feedback on a particular feature of the product. With this problem, the input data needs to be a set of products with the same set of features that the user is interested in and the sets of comments corresponding to each product. Each product in the set has the same features, it is included in the feature set. For each product is a set of customer reviews about the product. For each user comemnt, the system automatically extracts product information, features, phrases, and predicts the score for the product's performance with that opinion.

Keywords: Recommender systems, opinion mining, topic model, collaborative filtering, Content-based Filtering.