Presentation | 2020-02-14 A Data Fusion Method Assuming Latent Proxy Variables for Target Variables Yoshihide Nishio, Yasuo Tanida, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | We propose an analysis method that enables cross-domain prediction and interpretation of consumer behavior, and maintains anonymity in each domains sharing common consumer values. Such analysis has problems of data-sparsity. Our method solve it by compressing dimensions of consumer behavior while keeping relationship with ones values. We tests the method in 2-ways; the “upside” estimates consumer values diagnosed by a special questionnaire from purchased items in a supermarket, and the “downside” predicts from the opposite. It is worth noting that the downside keeps accuracy if input the upside estimation. We also shows an analysis on middle-layer of downside model that helps understanding customers. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Data Fusion / Consumer Behavior / Consumer Values / Cross-Domain Recommender / Matrix Factorization / Representation Learning |
Paper # | AI2019-52 |
Date of Issue | 2020-02-07 (AI) |
Conference Information | |
Committee | AI |
---|---|
Conference Date | 2020/2/14(1days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | Izumo Campus, Shimane University |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | Socionetwork Strategies in the Market of Data VI: Cross-cultural Collaboration and Life Space Innovation, etc. |
Chair | Naoki Fukuta(Shizuoka Univ.) |
Vice Chair | Yuichi Sei(Univ. of Electro-Comm.) / Yuko Sakurai(AIST) |
Secretary | Yuichi Sei(Osaka Univ.) / Yuko Sakurai(Tokyo Univ. of Agriculture and Technology) |
Assistant |
Paper Information | |
Registration To | Technical Committee on Artificial Intelligence and Knowledge-Based Processing |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | A Data Fusion Method Assuming Latent Proxy Variables for Target Variables |
Sub Title (in English) | |
Keyword(1) | Data Fusion |
Keyword(2) | Consumer Behavior |
Keyword(3) | Consumer Values |
Keyword(4) | Cross-Domain Recommender |
Keyword(5) | Matrix Factorization |
Keyword(6) | Representation Learning |
1st Author's Name | Yoshihide Nishio |
1st Author's Affiliation | Synergy Marketing, Inc.(Synergy Marketing) |
2nd Author's Name | Yasuo Tanida |
2nd Author's Affiliation | Synergy Marketing, Inc.(Synergy Marketing) |
Date | 2020-02-14 |
Paper # | AI2019-52 |
Volume (vol) | vol.119 |
Number (no) | AI-413 |
Page | pp.pp.55-60(AI), |
#Pages | 6 |
Date of Issue | 2020-02-07 (AI) |