Un Lee two , Sungwon Kang three, , Jongsun Ahn 3 and Heetae Cho2Department

Un Lee two , Sungwon Kang three, , Jongsun Ahn 3 and Heetae Cho2Department of AI Convergence Engineering (Graduate) and Aerospace and Software Engineering (Undergraduate), Gyeongsang National University, Jinju 52828, Korea; [email protected] (S.L.); [email protected] (H.C.) BigPictureLabs Inc., Daejeon 34047, Korea; [email protected] School of Computing, Korea Advanced Institute of Science and Technology (KAIST), Daejeon 34141, Korea; [email protected] Correspondence: [email protected]; Tel.: +82-42-350-Abstract: When performing computer software evolution tasks, Bevantolol Epigenetic Reader Domain developers D-Fructose-6-phosphate (disodium) salt Purity & Documentation commit a important quantity of time searching for files to modify. By recommending files to modify, a code edit recommendation method reduces the developer’s navigation time when conducting application evolution tasks. Within this paper, we propose a code edit recommendation method making use of a recurrent neural network (CERNN). CERNN forms contexts that preserve the sequence of developers’ interactions to advise files to edit and stops suggestions when the very first recommendation becomes incorrect for the given evolution task. We evaluated our method by comparing it with the state-of-the-art method MI-EA that was created depending on the association rule mining strategy. The result shows that our proposed process improves the typical recommendation accuracy by about 5 over MI-EA (0.64 vs. 0.59 F-score). Keyword phrases: data-based software program engineering; code edit recommendation; recurrent neural network; machine mastering; interaction historiesCitation: Lee, S.; Lee, J.; Kang, S.; Ahn, J.; Cho, H. Code Edit Recommendation Employing a Recurrent Neural Network. Appl. Sci. 2021, 11, 9286. https://doi.org/10.3390/ app11199286 Received: 6 September 2021 Accepted: 30 September 2021 Published: 6 October1. Introduction Developers commit a substantial amount of time on the lookout for files to modify when performing software evolution tasks [1]. By recommending files to modify [2], a code edit recommendation method allows developers to reduce search time through computer software evolution tasks. To reduce the time spent by the developer on code navigation, Zimmermann et al. [3] created a recommendation technique for mining association rules amongst changed files by collecting the revision history stored in the version management technique. The recommendation technique ROSE recommends files to edit using the context of one changed file and yields a 0.33 F-score [3]. Subsequently, Lee et al. [1] developed a recommendation technique MI-EA that functions by mining association rules amongst viewed files and edited files from an interaction history, such as what was recorded by Mylyn [4] and stored within the Eclipse Bugzilla program. MI-EA recommends files to edit with all the context of one particular edited file and 3 viewed files and yields a 0.58 F-score [1]. Lee et al. [1] viewed revision history as a subset from the interaction history in that revision history utilizes edit data for suggestions whereas interaction history utilizes each view information and edit information and facts for recommendations. Lee et al. [1] also showed that Zimmerman’s system applied to the interaction history obtains practically the exact same accuracy values as Zimmerman’s [3]. In line with Lee et al. [1], MI-EA yields a recommendation accuracy that is certainly higher than that of ROSE since MI-EA utilizes a a lot more elaborate context that contains viewed files. Both of the recommendation systems of Lee et al. [1] and Zimmermann et al. [3] use association rule mining as the key strategy. Alt.