| Project/Area Number |
18K18159
|
|---|
| Research Category |
Grant-in-Aid for Early-Career Scientists
|
| Allocation Type | Multi-year Fund |
|---|
| Review Section |
Basic Section 62020:Web informatics and service informatics-related
|
|---|
| Research Institution | The University of Electro-Communications |
|---|
Principal Investigator |
Okamoto Kazushi 電気通信大学, 大学院情報理工学研究科, 准教授 (10615032)
|
|---|
| Project Period (FY) |
2018-04-01 – 2022-03-31
|
| Project Status |
Completed (Fiscal Year 2021)
|
| Budget Amount *help |
¥4,160,000 (Direct Cost: ¥3,200,000、Indirect Cost: ¥960,000)
Fiscal Year 2020: ¥1,170,000 (Direct Cost: ¥900,000、Indirect Cost: ¥270,000)
Fiscal Year 2019: ¥780,000 (Direct Cost: ¥600,000、Indirect Cost: ¥180,000)
Fiscal Year 2018: ¥2,210,000 (Direct Cost: ¥1,700,000、Indirect Cost: ¥510,000)
|
|---|
| Keywords | 情報推薦 / 推薦理由の説明 / 協調フィルタリング / 線形回帰モデル / 図書推薦システム / 研究者推薦システム / 次元削減 / ニューラルネットワーク / 回帰分析 |
|---|
| Outline of Final Research Achievements |
This study develops model-based collaborative filtering techniques with explanations for recommended items based on linear regression. The proposed model treats users and items simultaneously in a single linear regression equation, and the reason for recommendation can be explained by presenting regression coefficients. We evaluated that the proposed model has achieved 25 times faster and the same level of recommendation accuracy, compared with Factorization Machines. In addition, a linear regression model with interaction between users and items is proposed. According to the results, we validated that the linear regression models have potentials for fast and accurate enough to make recommendations.
|
| Academic Significance and Societal Importance of the Research Achievements |
協調フィルタリングの実現手法にはメモリベース法とモデルベース法の２つがあり，推薦理由の説明に関する多くの研究はメモリベース法を対象としている．一方で，モデルベース法は推薦モデルが学習できれば高速な推薦を行えるが，既存手法では推薦がどの変数に影響を受けたかの把握が難しく，推薦理由の説明には直接活用できない課題がある．本研究の意義は，モデルベース協調フィルタリングにおいて，線形回帰モデルという予測の透明性を有し広く知られた手法を用いることでも高速かつ十分な精度の推薦が行えることを明らかにしたところにある．
|
