نوع مقاله : علمی-پژوهشی
نویسندگان
1 مدرس دانشگاه لرستان
2 دانشگاه اراک
3 گروه کامپیوتر، دانشگاه ملایر، ملایر
چکیده
کلیدواژهها
موضوعات
عنوان مقاله [English]
نویسندگان [English]
Nowadays, application developers use software bug detection techniques widely. One of the most popular bug detection techniques is static analysis which is a pattern-based method. In such techniques patterns are manually created by experts. Despite creating a huge amount of patterns for various bugs, there are still many bugs that pass through all the available filters. In this paper, a new approach is presented for automatic bug detection in JavaScript codes. We map the buggy and bug-free codes to graphs. Then a deep learning model which accepts graphs is trained to classify codes to buggy and bug-free. The evaluation results show that the proposed method covers a wider range of bugs while outperforms previous methods.
کلیدواژهها [English]
)