Collaborative filtering using dual information sources

Collaborative filtering (CF) is one of the most successful recommendation technique, which has proved its usefulness in various e-commerce applications. Using CF, a new method named Dual Information Source Model-Based Collaborative Recommender System (DISCORS) has been developed and applied to a high involvement product. There are two views of collaborative filtering and they include similarity-based CF and trust-based CF. The DISCORS recommendation process combines offline mining and online recommendation by creating user rating profiles, and extracting source receptivity. It generates personalized recommendations in real time by combining each user's source receptivity values with each information source's prediction terms. DISCORS uses the prediction terms to extract source receptivity and generate recommendations. It recommends an item when the item's predicted rating is greater than 70 percent of the maximum preference value.