A knowledge-based product recommendation system for e-commerce

This paper presents a knowledge-based product recommendation system for Business-to-Customer (B2C) e-commerce purposes. The system is based on the observation that the purchase patterns of previous users play a vital role in recommending the products to new users if the new users already followed parts of the existing patterns. The system is based on Case-Based Reasoning Plan Recognition (CBRPR) approaches and Automated Collaborative Filtering (ACF) approaches. The system also addresses the issue of organising and utilising the information related to the products that are purchased repetitively by a user. The system is named RecommendEx and is tested in a simulated environment to test its operational performance. The evaluation results are included.