A Regression-Based Approach for Scaling-Up Per- sonalized Recommender Systems in E-Commerce
Slobodan Vučetić, Zoran Obradović · 2000
Automated collaborative filtering is one of the key techniques for providing a customization for Ecommerce sites. Various neighbor-based recommendation methods are popular choices for collaborative filtering. However, their latency can be a serious drawback for scaling up to a large number of requests that should be processed in real-time. In this paper we propose an alternative regression-based approach that searches for relationships among items instead of looking for similarities among users. Experiments on a movie database provide evidence that the proposed regression-based approach provides significantly better accuracy and is two orders of magnitude faster than the neighbor-based alternatives. Even faster time response with accuracy similar to neighbor-based recommendations was obtained by adjusting the generic recommendations with only the average preference of an active user. 1. INTRODUCTION In today's society there is an increasing need for automated systems providing person...