A Collaborative Filtering-based Product Recommender in Electronic Commerce

다국어 초록 (Multilingual Abstract)

Collaborative filtering has been known to be the most successful recommendation technique that has been used in a number of different applications As both the number of customers and the number of products managed in an e-commerce site grow rapidly, however. its widespread use in e-commerce has exposed two major issues that must be addressed The first issue is to reduce the sparsity for the better quality of recommendations and the second issue is to improve the scalability for the better system performance In this paper, we propose a recommendation methodology based on Web usage mining and the product taxonomy to address these issues Web usage mining populates the rating database by tracking the customer shopping behavior on the Web, so results in overcoming the sparsity problem The product taxonomy is used both to reduce the sparsity of ratings and to improve the scalability of searching for like-mined customers through dimensionality reduction of the rating database We experimentally evaluate our methodology on real edata and compare them to the nearest neighbor algorithm Experimental results show that our methodology provides higher quality recommendations and better performance than the nearest neighbor algorithm.

목차 (Table of Contents)

• Abstract
  
• 1. Introduction
  
• 2. Methodology
  
• 3. Experimental Evaluation
  
• 4. Conclusion
  
• References