Title: A personalised recommendation procedure based on dimensionality reduction and web mining

Authors: Do Hyun Ahn, Jae Kyeong Kim, Il Young Choi, Yoon Ho Cho

Addresses: School of Business Administration, Kyung Hee University, #1, Hoeki-dong, Dongdaemoon-gu, Seoul, 130-701, South Korea. ' School of Business Administration, Kyung Hee University, #1, Hoeki-dong, Dongdaemoon-gu, Seoul, 130-701, South Korea. ' School of Business Administration, Kyung Hee University, #1, Hoeki-dong, Dongdaemoon-gu, Seoul, 130-701, South Korea. ' School of e-Business, Kookmin University, Seoul, South Korea

Abstract: With the advance of e-commerce, the importance of recommender systems grows larger than before. Collaborative filtering is the most successful recommendation method, but its widespread use has exposed some problems such as sparsity and scalability. To overcome such limitations, this paper proposes hybrid recommendation methodologies based on web usage mining and dimensionality reduction techniques to enhance the recommendation quality and performance of current collaborative filtering-based recommender systems. Web usage mining populates the rating database by tracking customers| shopping behaviours on the web, thereby leading to better quality recommendations. The product taxonomy and SVD (Singular Value Decomposition) are used to improve the performance of searching for nearest neighbours through dimensionality reduction of the rating database. Experiments on real e-commerce data show that the proposed methodologies provide higher quality recommendations and better performance than existing collaborative filtering methodologies.

Keywords: personalised recommendation; collaborative filtering; web usage mining; SVD; product taxonomy; singular value decomposition; dimensionality reduction; e-commerce; electronic commerce.

DOI: 10.1504/IJIEM.2004.005372

International Journal of Internet and Enterprise Management, 2004 Vol.2 No.3, pp.280 - 298

Published online: 29 Sep 2004 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article