Sorting and Transforming Program Repair Ingredients via Deep Learning Code Similarities
This work addresses a bottleneck in automated program repair for software maintainers, but it is incremental as it builds on existing redundancy-based techniques without major performance gains.
The authors tackled the problem of prioritizing and transforming code for automated program repair by using deep learning to reason about code similarities, resulting in DeepRepair finding patches faster but not more effectively than a baseline in terms of patch counts or attempts.
In the field of automated program repair, the redundancy assumption claims large programs contain the seeds of their own repair. However, most redundancy-based program repair techniques do not reason about the repair ingredients---the code that is reused to craft a patch. We aim to reason about the repair ingredients by using code similarities to prioritize and transform statements in a codebase for patch generation. Our approach, DeepRepair, relies on deep learning to reason about code similarities. Code fragments at well-defined levels of granularity in a codebase can be sorted according to their similarity to suspicious elements (i.e., code elements that contain suspicious statements) and statements can be transformed by mapping out-of-scope identifiers to similar identifiers in scope. We examined these new search strategies for patch generation with respect to effectiveness from the viewpoint of a software maintainer. Our comparative experiments were executed on six open-source Java projects including 374 buggy program revisions and consisted of 19,949 trials spanning 2,616 days of computation time. DeepRepair's search strategy using code similarities generally found compilable ingredients faster than the baseline, jGenProg, but this improvement neither yielded test-adequate patches in fewer attempts (on average) nor found significantly more patches than the baseline. Although the patch counts were not statistically different, there were notable differences between the nature of DeepRepair patches and baseline patches. The results demonstrate that our learning-based approach finds patches that cannot be found by existing redundancy-based repair techniques.