DeepSumm -- Deep Code Summaries using Neural Transformer Architecture
This addresses the need for efficient and scalable code documentation to aid software development and maintenance, though it is incremental as it adapts an existing architecture to a new domain.
The paper tackles the problem of automatically generating natural language summaries for source code, which is costly to do manually, by comparing neural machine translation (NMT) techniques to a Transformer architecture on a dataset of Java methods and comments. It reduces training time by over 50% and achieves a BLEU score of 17.99 on the test set.
Source code summarizing is a task of writing short, natural language descriptions of source code behavior during run time. Such summaries are extremely useful for software development and maintenance but are expensive to manually author,hence it is done for small fraction of the code that is produced and is often ignored. Automatic code documentation can possibly solve this at a low cost. This is thus an emerging research field with further applications to program comprehension, and software maintenance. Traditional methods often relied on cognitive models that were built in the form of templates and by heuristics and had varying degree of adoption by the developer community. But with recent advancements, end to end data-driven approaches based on neural techniques have largely overtaken the traditional techniques. Much of the current landscape employs neural translation based architectures with recurrence and attention which is resource and time intensive training procedure. In this paper, we employ neural techniques to solve the task of source code summarizing and specifically compare NMT based techniques to more simplified and appealing Transformer architecture on a dataset of Java methods and comments. We bring forth an argument to dispense the need of recurrence in the training procedure. To the best of our knowledge, transformer based models have not been used for the task before. With supervised samples of more than 2.1m comments and code, we reduce the training time by more than 50% and achieve the BLEU score of 17.99 for the test set of examples.