Learning-based automated machine learning code annotation with graph neural network

This invention describes a computer program that automatically writes comments for segments of computer code. It works by first identifying which stage of a data science project the code belongs to (like preparing data or building a model) and what kind of comment is needed (such as explaining the process or the result). Then, it generates a natural language explanation based on the code and the identified comment type, displaying it next to the code. The specific types of comments are limited to process, result, education, reference, and reason, and the project stages are limited to environment configuration, data preparation, feature engineering, and model building.