個人の感性モデルを推定する商品推薦システム

概要

論文の詳細を見る
In this paper, we proposed a recommendation method using user's personal Kansei model, which was estimated by interactive Genetic Algorithm (iGHA). When processing contents, this method extracts words which are representing the contents, and assigns these extracted words as content parameters. Then, this method constructs a contents parameter network in which the distance between nodes is defined by the similarities between them. By searching on a design variables space based on the contents parameter network, iGA estimates a user's Kansei model and recommends contents which are considered to be suitable for the user. In the experiment, the products recommended to a subject using the proposed method had the keywords which were similar to the characteristic of the products that he or she had already selected. This result indicated that the proposed method executed the searches properly, and obtained the contents which fitted his or her Kansei model.
同志社大学の論文
2013-10-00