Deep Neural Embedding for Software Vulnerability Discovery: Comparison and Optimization

Table 3

Reviewed studies which applied various embedding techniques for software engineering.

Paper	Type of data	Embedding model	Whether to consider contextual information
Pradel and Sen [23]	150,000 JavaScript files collected from various open-source projects	Word2Vec	No
Harer et al. [13]	C/C++ packages distributed with the Debian Linux distribution C/C++ functions collected from github	Word2Vec	No
Henkel et al. [24]	19,000 API-usage analogies extracted from the Linux kernel	GloVe	No
Fang et al. [25]	Projects are extracted from open-source intelligence data such as NVD	FastText	No
Kanade et al. [26]	150k Python files from github	CuBERT	Yes
Karampatsis and Sutton [27]	150,000 JavaScript files consisting of various open-source projects	SCELMo	Yes