개인화 상품 추천이 구매의도에 미치는 영향에 관한 연구 : Matrix Factorization 알고리즘을 중심으로 = A Study on the Effect of Personalized Product Recommendation on Purchasing Intention : Focused on Matrix Factorization algorithm

다국어 초록 (Multilingual Abstract)

The role of product recommendation as well as direct search associated with online shopping becomes increasingly important. The method of recommendation also seeks to provide personalized recommendations that fit the customer's tendency by actively reflecting and scoring the customer's behavioral history, away from the methods previously performed by humans. In general, the collaborative filtering method, which is frequently used in recommendation, causes a problem of sparsity when the usage history of the past is small, so this study proposed a personalized recommendation using the matrix factorization method. In order to verify the performance, A/B test was conducted by dividing into two groups that applied the existing merchandiser recommendation product and personalized recommendation product in email marketing. Through this, it was verified that the personalized recommendation increased the click conversion rate by 117.2% compared to the existing method. In the future, we will try to increase the efficiency of personalized email marketing by using segment information by gender and age, or recommending products related to shopping carts.

참고문헌 (Reference)

1 Goldberg, D., "Using collaborative filtering to weave an inf ormation tapestry" 35 (35): 61-70, 1992

2 Schafer, J. B., "The adaptive web" Springer 291-324, 2007

3 Schafer, J. B., "Recommender systems in e-commerce" 158-166, 1999

4 Takács, G., "Matrix factorization and neighbor based algorithms for the netflix prize problem" 267-274, 2008

5 Koren, Y., "Matrix factor ization techniques for recommender systems" 42 (42): 30-37, 2009

6 Sarwar, B., "Item-based collaborative filtering recommen dation algorithms" 285-295, 2001