Use of Similarity Measures in Collaborative Filtering Based on Binary User-Item Matrix

고객-제품 구매여부 데이터를 이용한 협동적 필터링에서의 유사성 척도의 사용

Collaborative filtering (CF) is originally based on the ratings of customers who vote on the items they used. When customers' votes are not available, user-item binary data set which represents choice and non-choice can also be used in this analysis. In this case the similarities between active user and the other users must be modified. Therefore we compare eight types of binary similarities by applying them in the modified CF Algorithm. Some experimental results will be reported.