About FABLE

FABLE (Finding Aliases for Broken Links Efficiently) is a system developed at USC that automatically discovers new URLs for broken links. The project addresses the widespread issue of "link rot" on the internet, where over 98% of web links break over time.

Research Impact

• Successfully tested on ~500,000 pages across Wikipedia, Medium and StackOverflow
• Found new URLs for 25% of broken links with 90% accuracy
• Currently working with Wikipedia to fix thousands of broken external links
• Published at ACM Internet Measurement Conference 2023

How It Works

FABLE leverages the fact that website reorganizations typically follow programmatic patterns. For example:

Old: http://ruby.railstutorials.org/chapters/following-users

New: https://www.railstutorial.org/book/following_users

The system identifies and reverse engineers these URL transformation patterns to locate moved content.

Team

Led by Professor Harsha Madhyastha at USC Computer Science with lead author Jingyuan Zhu from University of Michigan. Supported by the Alfred P. Sloan Foundation.