| Presentation | 2018-11-02 Method for Detecting Near-duplicate Recipes Based on Nearest Neighbor Search for Features of Cooking Instructions and Food Images Masaki Oguni, Lasguido Nio, Yu Hirate, Yohei Seki, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | In user-generated recipe websites, recipes that imitated recipe elements of other recipes are posted. In this study, we call these recipes as “near-duplicate recipes”. We propose the methods for detecting near-duplicate recipes based on cooking instructions similarities and number of differences of ingredients, and also based on food images similarities. We extracted cooking instruction features based on the distributed representation of words trained on using approximately 1.16 millions of recipes. In addition, we extracted food image features based on pre-trained Inception-v3. We utilize nearest neighbor search technique to reduce search time for near-duplicate recipes. We conducted the experiment to evaluate the proposed method. As a result, we confirmed that our proposed method were able to detect near-duplicate recipes, which the writer seemed to imitate the original recipe with paraphrasing the terms in the cooking instructions of it. We also measured execution time of our proposed method using nearest neighbor search and found that we could reduce execution time to approximately one seventieth compared to our previous method using exhaustive search. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Near-duplicate recipe / Nearest neighbor search / Similar text search / Similar image search |
| Paper # | HCS2018-44 |
| Date of Issue | 2018-10-26 (HCS) |
| Conference Information | |
| Committee | HCS |
|---|---|
| Conference Date | 2018/11/2(1days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | Masafumi Matsuda(NTT) |
| Vice Chair | Nobuyuki Watanabe(Kanazawa Inst. of Tech.) / Tomoo Inoue(Univ. of Tsukuba) |
| Secretary | Nobuyuki Watanabe(Ritsumeikan Univ.) / Tomoo Inoue(Osaka Electro-Comm. Univ.) |
| Assistant | Kazuki Takashima(Tohoku Univ.) / Ken Fujiwara(Osaka Univ. of Economic) / Kazunori Terada(Gifu Univ.) / Atsushi Kimura(Nihon Univ.) |
| Paper Information | |
| Registration To | Technical Committee on Human Communication Science |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Method for Detecting Near-duplicate Recipes Based on Nearest Neighbor Search for Features of Cooking Instructions and Food Images |
| Sub Title (in English) | |
| Keyword(1) | Near-duplicate recipe |
| Keyword(2) | Nearest neighbor search |
| Keyword(3) | Similar text search |
| Keyword(4) | Similar image search |
| 1st Author's Name | Masaki Oguni |
| 1st Author's Affiliation | University of Tsukuba(Tsukuba Univ.) |
| 2nd Author's Name | Lasguido Nio |
| 2nd Author's Affiliation | Rakuten Institute of Technology(RIT) |
| 3rd Author's Name | Yu Hirate |
| 3rd Author's Affiliation | Rakuten Institute of Technology(RIT) |
| 4th Author's Name | Yohei Seki |
| 4th Author's Affiliation | University of Tsukuba(Tsukuba Univ.) |
| Date | 2018-11-02 |
| Paper # | HCS2018-44 |
| Volume (vol) | vol.118 |
| Number (no) | HCS-278 |
| Page | pp.pp.19-24(HCS), |
| #Pages | 6 |
| Date of Issue | 2018-10-26 (HCS) |