Penanganan Spasity Untuk Meningkatkan Kualitas Rekomendasi Pada Collaborative Flittering

Abstract

The development of e-commerce platforms in this century has been very rapid and currently there are many e-commerce sites in Indonesia that sell various kinds of products and services. A recommendation system is a system for filtering, sorting items and information that takes preferences from user behavior, user profiles or opinions from the user community to assist individuals in identifying interesting and potentially high-potential content to select, purchase or use. In this study, data that continues to grow requires a recommendation as an alternative to refinding the items needed. This is because the resulting recommendations have better quality. Until this research tries to analyze sparsity handling.From the results of the evaluation, the conclusion that can be drawn is that the evaluation process results in a higher ranking of 1% Borda-count. Based on the results of the significance test, the Borda Algorithm runs around 28,000 times faster. In addition, the imputation stages carried out also affect the predicted ranking value. Meanwhile, based on accuracy metrics, the proposed method obtains 2 times higher Coverage values, 19% higher NDCG values.