Logo image
Generating comments from source with logical forms
Dissertation   Open access

Generating comments from source with logical forms

Sergey Matskevich
Doctor of Philosophy (Ph.D.), Drexel University
Mar 2024
DOI:
https://doi.org/10.17918/00001944
pdf
Matskevich_Sergey_2024671.60 kBDownloadView

Abstract

Information networks Form (Logic)
Code documentation and description is a big problem in the industry and many have attempted to automate comment generation. Earlier approaches with various success rates commonly tried to build comment as a function description based on the names of objects and functions in the code. Modern approaches are based on machine learning and the latest LLM-based comment generation often produce good results. However none of these approaches guarantee correctness of the generated text or even the correctness of English language in the sentence. I am proposing a different approach to generate comments, which will guarantee correctness and grammatically correct English. My approach is to generate description of an existing code base by treating programming language as a natural language and mapping to to logical expression and transforming PL into English.

Metrics

111 File views/ downloads
93 Record Views

Details

Logo image