Implementation of weighted parallel hybrid recommender systems for e-commerce in Indonesia

Abstract

This paper focus on building recommender system with weighted parallel hybrid method for e-commerce in Indonesia. The dataset was derived from one of the largest ecommerce company in Indonesia. The experiments used three sampling techniques, namely bootstrapping validation, timing series and systematic sampling. The best result of these experiments yields F1-measure of 9.99% which is achieved by the combination of user-based collaborative filtering approach and content-based filtering approach. Moreover, the value of evaluation metrics in this research is not much different from the previous research of recommender system. This indicates that recommender systems can be applied to e-commerce companies in Indonesia.