A Context-Aware Ranking Method based on Prediction of Important Parameters for a User
Kenta Oku, Hirokazu Kato · 2008
We propose a context-aware ranking method which can rank recommendation items appropriating user's current contexts. It is dicult to improve users' satisfaction for information recommendation by standard ranking methods since user's needs change according to the current contexts. In this study, we propose a novel ranking method which predicts what parameters are important for the user at each context from a user's preference model based on an SVM, and ranks recommendation items based on the important parameters. In this paper, we show eectiveness of our proposed method from experimental results.