Applying natural language processing techniques to code
Year: 2022
Type: dissertation
Abstract: Analyzing source code is becoming a helpful method as the amount of code and the number of coders increases. Developing software is a mentally strenuous activity because it involves understanding and formalizing solutions to complex problems. As solutions are adapted and the problem themselves change, the code changes with it. Traditionally, source code analysis (SCA) uses methods and tools that rely on the rigid structure of code to work, and while useful, are not always able to adapt to new information and code. By comparison, the natural language processing (NLP) field leverages naturally occurring patterns within natural language text to help people understand and interpret text robustly on a large scale. This works shows that methods from NLP can be applied to code to extract information automatically in broad and more adaptable ways. We show several examples of how techniques developed for NLP can be successfully applied to software to create source code tools that can help developers improve source code.
DOI: 10.17918/00001100
Authors: Aviel J. Stein, Spiros Mancoridis
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
Funder: Drexel University
OpenAlex ID: W4285731393
