期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2018
卷号：96
期号：14
出版社：Journal of Theoretical and Applied
摘要：E-commerce company delivered product information to customer or customer candidate trough web portal. There is basic mechanism which a system has belong responsible to calculate and predict information that suitable to customers or customers candidate interested namely recommender system. Most successful approach to calculate customer/user interest are based on collaborative filtering. This approach relies on rating from customers to products or items as a basic approach aims calculate similarity of users responds about items to produce recommendation. In fact, just a little number of customers who giving the rating approximately less than 1 percent from all customer population in datasets. It�s a reason of rising sparse data. In this research used 2 technical approach to deal with sparse data consist Non-Negative Matrix factorization to reduce dimensional reduction and involve deep learning to compute latent factor in a part of users, item and rating. This research consider dataset from MovieLens, many researchers believe to conduct experiment their approach algorithm to increase better performance. Final experiment we used RME (Root Mean Error) and RMSE (Root Mean Square Error) to measure accuracy of result experiment and according the result, our approach has obtained good result to reduce missing value.
关键词：Non-Negative; E-Commerce; Recommender System; Collaborative Filtering; Matrix Factorization; Deep Learning
