| Presentation | 2012-03-12 Detecting Latent Structural Changes via Latent Dirichlet Allocation Masashi UEDA, Ryota TOMIOKA, Kenji YAMANISHI, Katsuhiko ISHIGURO, Hiroshi SAWADA, Naonori UEDA, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | Detecting changes in consumers' latent preference is a fundamental challenge for improving recommendation systems as well as market analysis. In this paper, we propose a method for detecting changes in consumers' preference from purchase logs in an online setting. The proposed method employs a Latent Dirichlet Allocation (LDA) model that shares common hyperparameters before and after an assumed change point. The change in latent preference is measured as the distance between the two LDA distributions corresponding to before and after the assumed change point. Furthermore, a dynamic threshold optimization technique is used to raise alarms at change points based on the above distance. We create data sets with artificial change points by concatenating purchase logs from different consumers. We demonstrate on these data sets that the proposed method performs significantly better than a simple multinomial model that does not have latent variables. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Latent Dirichlet Allocation / Change point detection / user preference model / purchase logs |
| Paper # | IBISML2011-89 |
| Date of Issue |
| Conference Information | |
| Committee | IBISML |
|---|---|
| Conference Date | 2012/3/5(1days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | |
| Vice Chair | |
| Secretary | |
| Assistant | |
| Paper Information | |
| Registration To | Information-Based Induction Sciences and Machine Learning (IBISML) |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Detecting Latent Structural Changes via Latent Dirichlet Allocation |
| Sub Title (in English) | |
| Keyword(1) | Latent Dirichlet Allocation |
| Keyword(2) | Change point detection |
| Keyword(3) | user preference model |
| Keyword(4) | purchase logs |
| 1st Author's Name | Masashi UEDA |
| 1st Author's Affiliation | Faculty of Engineering, The University of Tokyo() |
| 2nd Author's Name | Ryota TOMIOKA |
| 2nd Author's Affiliation | Graduate School of Information Science and Technology, The University of Tokyo |
| 3rd Author's Name | Kenji YAMANISHI |
| 3rd Author's Affiliation | Graduate School of Information Science and Technology, The University of Tokyo |
| 4th Author's Name | Katsuhiko ISHIGURO |
| 4th Author's Affiliation | NTT Communication Science Laboratories, NTT Corporation |
| 5th Author's Name | Hiroshi SAWADA |
| 5th Author's Affiliation | NTT Communication Science Laboratories, NTT Corporation |
| 6th Author's Name | Naonori UEDA |
| 6th Author's Affiliation | NTT Communication Science Laboratories, NTT Corporation |
| Date | 2012-03-12 |
| Paper # | IBISML2011-89 |
| Volume (vol) | vol.111 |
| Number (no) | 480 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |