Predicting Code Comment Locations with Machine Learning

In this research project led by Annie Louis and co-authors, including researchers from esteemed institutions, a dataset and model were developed to predict optimal code comment locations. The study delves into the challenge of determining where to place comments in code efficiently to guide developers towards writing more useful comments. The project involves training a machine learning model using a corpus of C/C++ code with annotated comment locations sourced from professional projects.

• Machine Learning
• Code Comments
• Dataset
• Model
• C/C++

sgri Follow

Uploaded on Feb 18, 2025 | 2 Views

Download Presentation

Please find below an Image/Link to download the presentation.

The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author.If you encounter any issues during the download, it is possible that the publisher has removed the file from their server.

You are allowed to download the files provided on this website for personal or commercial use, subject to the condition that they are used lawfully. All files are the property of their respective owners.

The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author.

Presentation Transcript

1. Where should I comment my code? A dataset and model for predicting locations that need comments Annie Louis (University of Edinburgh, now works at Google) Co-authors: Santanu Kumar Dash (University of Surrey) Earl T. Barr (University College London) Michael D. Ernst (University of Washington) Charles Sutton (Google Research)

2. How many comments and where in the code? Too many comments use more developer time / possible comment rot Guide developers towards most useful locations A new research problem: Identify where to write comments A corpus of C/C++ code with comment locations A machine learning solution to identify these locations

3. Data for training a machine learning model Need a source of good commenting practice Android Open Source Project 601 files from 9 C/C++ libraries in the native substrate Professionally developed

4. Idea of Snippets code blocks delimited by empty lines 001 002 003 004 005 006 007 008 if (...) 009 010 /* Find Huffman .. */ if (...) ERREXIT1(...); htbl = isDC ? ... : Snippet 1 Label=1 comment here cinfo(...) ; Snippet 2 Label=0

5. Idea of Snippets code blocks delimited by empty lines 001 002 003 004 005 006 007 008 if (...) 009 010 /* Find Huffman .. */ if (...) ERREXIT1(...); htbl = isDC ? ... : Snippet 1 Label=1 comment here cinfo(...) ; Snippet 2 Label=0 Comments are removed!

6. Idea of Snippets code blocks delimited by empty lines 001 002 003 004 005 006 007 008 if (...) 009 010 /* Find Huffman .. */ if (...) ERREXIT1(...); htbl = isDC ? ... : Snippet 1 Label=1 comment here cinfo(...) ; Snippet 2 Label=0 Data = 41,506 snippets with binary labels

7. Best model: Hierarchical neural network

8. Best model: Hierarchical neural network Snippet 1 Snippet 2 L3 L4 L5 L6 L8 L9

9. Best model: Hierarchical neural network Snippet 1 Snippet 2 L3 L4 L5 L6 L8 L9 Recurrent neural network gets representations of lines of code based on prior lines hk hk+1 hi hi+1 ... ...

10. Best model: Hierarchical neural network Snippet 1 Snippet 2 L3 L4 L5 L6 L8 L9 hk hk+1 hi hi+1 ... ... Represent the sequence of snippets in the file jm jm+1

11. Best model: Hierarchical neural network Snippet 1 Snippet 2 L3 L4 L5 L6 L8 L9 hk hk+1 hi hi+1 ... ... jm jm+1 Obtain predictions 0 1

12. Performance on test data High precision to avoid incorrect suggestions Area for future work

13. Thank You Our code and data is available at http://groups.inf.ed.ac.uk/cup/comment-locator/