Where and what to comment

Abstract

Commenting is described as the process of placing Human Readable Descriptions within a com- puter program/code that summarizes the functionality of the code within which the description is written. The strength of a good comment lies in the fact that the user is not required to read the actual code to understand the working of the code. The project deals with identifying the locations as to where the comments are required and what should be the text that should be present in the commenting, all the while capturing the traits of informative comments. To achieve the same, the code is converted into graphical representations using language parsing tools and then the obtained graphs are processed with the help of Graph Neural Networks to achieve the desired target of location and the content of a comment. The dataset of this project is created from scratch. The results will be updated along the development of the project.