Tracking Customers' Preference Changes to Improve Accuracy of Item-Based Collaborative Filtering

As the number of products and services on the web increases, finding desired items by users becomes a challenge. Recommender systems are one of the tools trying to provide the personalized recommendations of items to users and solve this problem. One of the most common techniques in recommender systems is collaborative filtering, that despite its success and widespread use, has a problem yet; although the preferences of customers change over time existing collaborative filtering (CF) systems only incorporate rating information of users and consider ratings at different times equally. This may lead to recommendations based on old and changed preferences. To alleviate this problem a new recommendation methodology based on CF suggested which weights ratings dynamically according to current preferences of the user toward product categories. Experiments using real-world data demonstrate that the proposed methodology provides higher quality recommendations than do typical CF techniques