A systematic mapping study of source code representation for deep learning in software engineering

Abstract

The usage of deep learning (DL) approaches for software engineering has attracted much attention, particularly in source code modelling and analysis. However, in order to use DL, source code needs to be formatted to fit the expected input form of DL models. This problem is known as source code representation. Source code can be represented via different approaches, most importantly, the tree-based, token-based, and graph-based approaches. We use a systematic mapping study to investigate i detail the representation approaches adopted in 103 studies that use DL in the context of software engineering. Thus, studies are collected from 2014 to 2021 from 14 different journals and 27 conferences. We show that each way of representing source code can provide a different, yet orthogonal view of the same source code. Thus, different software engineering tasks might require different (combinations of) code representation approaches, depending on the nature and complexity of the task. Particularly, we show that it is crucial to define whether the DL approach requires lexical, syntactical, or semantic code information. Our analysis shows that a wide range of different representations and combinations of representations (hybrid representations) are used to solve a wide range of common software engineering problems. However, we also observe that current research does not generally attempt to transfer existing representations or models to other studies even though there are other contexts in which these representations and models may also be useful. We believe that there is potential for more reuse and the application of transfer learning when applying DL to software engineering tasks.

Software vulnerability detection method based on code attribute graph presentation and Bi-LSTM neural network extraction

“…This section mainly compares the time performance of VDCPG with the methods proposed by Lin et al [8], Jacob A et al [14], Canan Batur et al [15], X Cheng et al [16], SM Ghaffarian et al [17] and W Jian et al [18]. As can be seen from Table 8, Compared with other methods, the time cost of this method is mainly in the model pretreatment stage and training stage, and the time difference in the detection stage is little.…”

“…In the model pretreatment stage and the model training stage, VDCPG can directly input the code files in the project without segmentation, so the time required is shorter compared with other models. [8] 181 1041 35 Jacob A et al [14] 257 1124 54 Canan Batur et al [15] 149 1073 47 X Cheng et al [16] 381 975 59 SM Ghaffarian et al [17] 203 492 65 W Jian et al [18] 196 885 43 VDCPG 117 489 30…”