롱테일 상품의 일대일 마케팅 활성화를 위한 추천시스템에 관한 연구: 고객의 영화 선호도 예측에의 적용

The paper studies the long tail recommendation problem for the item-based collaborative filtering method and suggest the new recommendation method for the tail products. In internet era, the power of the hit items are decreased, on the other hand, the power of the tail items are increased. Thus, recommending the right products in the tail part as well as the head become very important. Recommender Systems have been used as successful 1-to-1 marketing tools, however they often ignore the unpopular items having only few rating. In this research we suggest the new recommendation technique works well both the head and the tail. The suggested method splits the whole products into the head and the tail parts and applies item-to-item collaborative filtering method in the head parts, however apply the new hybrid dataming technique using clustering and regression in the tail part. We apply this method to the real movie dataset. The result shows that the suggested method reduces the recommendation error rates for the tail items, while maintaining reasonable computational performance.