Tag based Product Recommendation System Using Rating Variance

Abstract

Information overload have become a major problem due to sudden rise in the Internet data and users visiting websites. The recommendation system is a technique of information filtering to predict the preferences that users may like. The main target of recommendation system is to provide useful information to users and solve the large-scale problem of information overload where users are not able to get correct results. Different products have different ratings given by many users. On most websites, each product has some labels or phrases, which are used to identify the content of a product called as tags. In practice, users generally look at the ratings and tags attached to the item to decide whether they should buy it or not. The proposed approach generates recommendations using collaborative filtering by utilizing users’ tags and ratings information. The proposed approach predicts rating of products to the target user based on target user’s profile and other similar users. The proposed approach adjusts the predicted ratings of recommended items using standard deviation of items’ ratings.