Extended Bayesian Model for Multi-criteria Recommender System
Search this article
Abstract
We have proposed multi-criteria (MC) recommender system by using a latent probabilistic model. In this model, users and items are mapped into small number of groups, and preference is represented based on the group instead of indivisual user. In other words, features of users and items are represented by probability distributions over latent topics. When predicting rating scores, we need to aggregate features into predicted rating score. This paper compares two ways to aggregate features for predicting rating score of unrated items in MC recommendation.We have proposed multi-criteria (MC) recommender system by using a latent probabilistic model. In this model, users and items are mapped into small number of groups, and preference is represented based on the group instead of indivisual user. In other words, features of users and items are represented by probability distributions over latent topics. When predicting rating scores, we need to aggregate features into predicted rating score. This paper compares two ways to aggregate features for predicting rating score of unrated items in MC recommendation.
Journal
-
- 研究報告情報基礎とアクセス技術(IFAT)
-
研究報告情報基礎とアクセス技術(IFAT) 2013 (6), 1-4, 2013-01-04
- Tweet
Details 詳細情報について
-
- CRID
- 1571417127824792704
-
- NII Article ID
- 110009495393
-
- NII Book ID
- AN10114171
-
- Text Lang
- en
-
- Data Source
-
- CiNii Articles