Skip to Main Content
In order to tackle the problems of current recommendation system, such as lack of domain knowledge and limited applications in e-commerce field, a domain-driven recommendation method based on binary data is proposed in this paper. By introducing the importance of domain knowledge and its application in recommendation, we discuss how to combine domain knowledge with collaborative filtering and apply it on binary purchasing data. Empirical result on supermarket dataset shows that domain-driven recommendation method outperforms other recommendation methods on three common evaluating indicators.
Date of Conference: 15-19 April 2011