Naturally!: How Breakthroughs in Natural Language Processing Can Dramatically Help Developers
Authors:
- Anand Ashok SawantASAnand Ashok Sawant
- Premkumar DevanbuPDPremkumar Devanbu
- 20 August 2021
- journal article
- editorial
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Software
- Vol. 38 (5) , 118-123
Abstract
Taking advantage of the naturalness hypothesis for code, recent development, and research has focused on applying machine learning (ML) techniques originally developed for natural language processing (NLP) to drive a new wave of tools and applications aimed specifically for software engineering (SE) tasks. This drive to apply ML and deep learning (DL) has been animated by the large-scale availability of software development data (e.g., source code, code comments, code review comments, commit data, and so on) available from open source platforms such as GitHub and Bitbucket.Keywords
This publication has 10 references indexed in Scilit:
- Learning lenient parsing & typing via indirect supervisionEmpirical Software Engineering, 2021
- Unified Pre-training for Program Understanding and Generation2021
- Comprehension of computer code relies primarily on domain-general executive brain regionseLife, 2020
- Do Programmers Prefer Predictable Expressions in Code?Cognitive Science, 2020
- Open-vocabulary models for source code2020
- A theory of dual channel constraints2020
- CodeBERT: A Pre-Trained Model for Programming and Natural Languages2020
- NL2Type: Inferring JavaScript Function Types from Natural Language Information2019
- Studying the difference between natural and programming language corporaEmpirical Software Engineering, 2019
- code2vec: learning distributed representations of codeProceedings of the ACM on Programming Languages, 2019