| Tweet | |
| J. Hong, E. Choi, and O. Mizuno, "A Combined Alignment Model for Code Search," IEICE Transactions on Information and Systems, E107-D(3), pp. 257-267, March 2024. | |
| ID | 944 |
| 分類 | 学術論文誌(査読付) |
| タグ | alignment code combined model search |
| 表題 (title) |
A Combined Alignment Model for Code Search |
| 表題 (英文) |
|
| 著者名 (author) |
Juntong Hong,Eunjong Choi,Osamu Mizuno |
| 英文著者名 (author) |
Juntong Hong,Eunjong Choi,Osamu Mizuno |
| キー (key) |
Juntong Hong,Eunjong Choi,Osamu Mizuno |
| 定期刊行物名 (journal) |
IEICE Transactions on Information and Systems |
| 定期刊行物名 (英文) |
|
| 巻数 (volume) |
E107-D |
| 号数 (number) |
3 |
| ページ範囲 (pages) |
257-267 |
| 刊行月 (month) |
3 |
| 出版年 (year) |
2024 |
| Impact Factor (JCR) |
|
| URL |
|
| 付加情報 (note) |
|
| 注釈 (annote) |
|
| 内容梗概 (abstract) |
Code search is a task to retrieve the most relevant code given a natural language query. Several recent studies proposed deep learn- ing based methods use multi-encoder model to parse code into multi-field to represent code. These methods enhance the performance of the model by distinguish between similar codes and utilizing a relation matrix to bridge the code and query. However, these models require more computational resources and parameters than single-encoder models. Furthermore, utilizing the relation matrix that solely relies on max-pooling disregards the delivery of word alignment information. To alleviate these problems, we propose a combined alignment model for code search. We concatenate the multi-code fields into one sequence to represent code and use one encoding model to encode code features. Moreover, we transform the relation matrix using trainable vectors to avoid information losses. Then, we combine intra- modal and cross-modal attention to assign the salient words while matching the corresponding code and query. Finally, we apply the attention weight to code/query embedding and compute the cosine similarity. To evaluate the performance of our model, we compare our model with six previous models on two popular datasets. The results show that our model achieves 0.614 and 0.687 Top@1 performance, outperforming the best comparison models by 12.2% and 9.3%, respectively. |
| 論文電子ファイル | 利用できません． |
| BiBTeXエントリ |
@article{id944,
title = {A Combined Alignment Model For Code Search},
author = {Juntong Hong and Eunjong Choi and Osamu Mizuno},
journal = {IEICE Transactions on Information and Systems},
volume = {E107-D},
number = {3},
pages = {257-267},
month = {3},
year = {2024},
}
|