A Study of Weight-Based Techniques for Recommender System Application
The International Journal of Engineering and Science
This study proposed popularity and awareness-based weight techniques which can be used in collaborative filtering approach. In this concept, relative popularity change and features which change depending on users' awareness ratio are applied as they are. Therefore, it is recommendable according to trends. Furthermore, it can improve a probability of being satisfied when a user selects a product after its relative quality is considered for recommendation. This study proposes a method which
... method which increases product quality with weight. This applied weight to a 2-D recommendation space which is evaluated by the recommender system operated in conventional web environment for recommendation, making recommendation considering multidimensional elements possible. MAEs on five experimental data were compared. The NBCFA which adopted product quality as weight revealed high recommendation accuracy. When popularity weight is used, NBCFA and CMA reveal 1.4% and 4.3% respectively in performance improvement.