Malinda Dilhara retweetledi

At @IEEEICSME, @danny_dig and @BellurAbhiram just presented our new paper "Next-Generation Refactoring: Combining LLM Insights and IDE Capabilities for Extract Method". 🎉
This work continues our long-time research into how to improve refactorings in the IDE using machine learning. Our approach uses LLMs to generate Extract Method suggestions. It then removes the ones that are invalid or not useful, and enhances the suggestions using static analysis and program slicing. On a diverse test-set of 1752 conducted EM refactorings, our plugin reached a recall@5 of 53.4%, significantly outperforming the existing state of the art. 💪
The tool is freely available for IntelliJ IDEA as a plugin. Congratulations to Dorin, @BellurAbhiram, @Malindadoo, @zkurbatova, @egor_bb, @timofeybryksin, and @danny_dig!
📜 Pre-print: danny.cs.colorado.edu/papers/EM-Assi…
🛠️ Plugin: plugins.jetbrains.com/plugin/23403-l…


English


















