| Presentation | 2022-03-07 Project-domain adaptation with RoBERTa Model for Code Completion Daisuke Fukumoto, Toshiki Hirao, Kenji Fujiwara, Hajimu Iida, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | Code completion is a function that automatically recommends code for developers when they are writing code. Code completion can reduce coding time even for developers with insufficient skills. In the previous research, the machine learning-based completion that considers the context of the program and completes tokens in a more generally has been proposed. However, these previous studies have reported that the accuracy of completion decreases rapidly when the number of completing tokens increases. In this study, we propose a method that additionally trains RoBERTa, a machine learning model for natural language processing, on a dataset built in the domain of a developer's repository. We expect it can reduce the vocabulary recommended by the model and improve the completion accuracy. As a result of our experiments, we confirmed a improvement in accuracy compared to pre-trained model. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Code Completion / Transfer Learning / BERT / RoBERTa |
| Paper # | SS2021-50 |
| Date of Issue | 2022-02-28 (SS) |
| Conference Information | |
| Committee | SS |
|---|---|
| Conference Date | 2022/3/7(2days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | Online |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | Software Science etc. |
| Chair | Takashi Kobayashi(Tokyo Inst. of Tech.) |
| Vice Chair | Kozo Okano(Shinshu Univ.) |
| Secretary | Kozo Okano(Hiroshima City Univ.) |
| Assistant | Shinpei Ogata(Shinshu Univ.) |
| Paper Information | |
| Registration To | Technical Committee on Software Science |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Project-domain adaptation with RoBERTa Model for Code Completion |
| Sub Title (in English) | |
| Keyword(1) | Code Completion |
| Keyword(2) | Transfer Learning |
| Keyword(3) | BERT |
| Keyword(4) | RoBERTa |
| 1st Author's Name | Daisuke Fukumoto |
| 1st Author's Affiliation | Nara Institute of Science and Technology(NAIST) |
| 2nd Author's Name | Toshiki Hirao |
| 2nd Author's Affiliation | Nara Institute of Science and Technology(NAIST) |
| 3rd Author's Name | Kenji Fujiwara |
| 3rd Author's Affiliation | Tokyo City University(TCU) |
| 4th Author's Name | Hajimu Iida |
| 4th Author's Affiliation | Nara Institute of Science and Technology(NAIST) |
| Date | 2022-03-07 |
| Paper # | SS2021-50 |
| Volume (vol) | vol.121 |
| Number (no) | SS-416 |
| Page | pp.pp.49-53(SS), |
| #Pages | 5 |
| Date of Issue | 2022-02-28 (SS) |