Towards Demystifying Dimensions of Source Code Embeddings
This work addresses the need for principled analysis of code representations for researchers and practitioners in software engineering and AI, but it is incremental as it presents preliminary results in a limited case study.
The paper tackled the problem of understanding the contents and characteristics of neural source code embeddings, specifically code2vec, by comparing them with handcrafted features in a small case study using binary SVM classifiers. The results showed that handcrafted features performed very close to high-dimensional code2vec embeddings, with information gains more evenly distributed in embeddings and greater resilience to dimension removal.
Source code representations are key in applying machine learning techniques for processing and analyzing programs. A popular approach in representing source code is neural source code embeddings that represents programs with high-dimensional vectors computed by training deep neural networks on a large volume of programs. Although successful, there is little known about the contents of these vectors and their characteristics. In this paper, we present our preliminary results towards better understanding the contents of code2vec neural source code embeddings. In particular, in a small case study, we use the code2vec embeddings to create binary SVM classifiers and compare their performance with the handcrafted features. Our results suggest that the handcrafted features can perform very close to the highly-dimensional code2vec embeddings, and the information gains are more evenly distributed in the code2vec embeddings compared to the handcrafted features. We also find that the code2vec embeddings are more resilient to the removal of dimensions with low information gains than the handcrafted features. We hope our results serve a stepping stone toward principled analysis and evaluation of these code representations.