Key points are not available for this paper at this time.
Debugging a program is always an obstacle to programmers and learners. In particular, novice programmers waste a lot of time finding bugs, so a feedback system to support debugging is required. Although existing editors and IDEs support finding syntax errors, their functions for detecting logical errors are limited. In the present paper, we present bug detection methods for the feedback system of an online judge system which contains many programming problems and accumulates numerous lines of solution source code. The proposed method uses the solutions and a language model based on long short-term memory (LSTM) networks for bug detection. In addition, since LSTM networks have some hyperparameters, we investigate the best model for bug detection in terms of perplexity and training time. The results of experiments show that models trained by solutions can detect bugs in a compiled code based on the static structure of a program.
Teshima et al. (Mon,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: