.png){kind=icon}
.png){kind=icon}
Return
Improving code search by query reformulation with experienced programmer intelligence
.png){kind=icon}
PRE
AI
DOI：10.1016/j.infsof.2026.108174.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Query expansion or reformulation involves enhancing short queries by replacing words, adding new terms, and other techniques to improve search results. In code search, a wealth of high-quality history search data are available from professional users. These data capture both the expressions used by the query creators and the content of their queries, providing valuable reference for the reformulation of queries for common users. In this study, we explore how to apply high-quality historical queries by experienced programmer to reconstruct the query submitted by common users, thereby enhancing the accuracy of their code search. To achieve this goal, we collect functional descriptions from GitHub and comment data from StackOverflow to compile a high-quality query dataset with 2 million descriptions from IT professionals. Based on the compiled dataset, a query reformulation method based on these professional description, named QRPD, is proposed to improve the accuracy of code search. Extensive experiments are conducted and experimental results demonstrate that QRPD outperforms the state-of-the-art query reformulation methods in code search with the accuracy being improved by 8%, which verifies the effectiveness and superiority of the proposed method.
Keywords:
query reformulation
code search
experienced programmer intelligence
query expansion
software engineering
Journal
IF:
4.3
Papers:
3.7K
Citations:
7.7K