How to programmatically spot errors
🔥 Learn how to enhance your data labeling process with programmatic data labeling QA. This video guides you through the process of programmatically spotting errors, a crucial step in maintaining high-quality data labeling. 🔎
✨ We delve into the use of Python rules for regular expression and field validation checks on entities extracted from a document. Discover how to create custom issues when errors are detected, and how to integrate these rules as a Kili plugin for automatic triggering upon asset submission by a labeler. 🤔
Watch as we demonstrate the use of our labeling interface to annotate an invoice, and see how our programmatic data labeling QA system identifies and flags errors. Learn how to filter assets with open issues in the Explore page, and how to correct errors and mark assets as reviewed. ✅
😍 With our programmatic data labeling QA, you can decide whether an asset should be sent back to the queue, corrected by the labeler, or added to review using custom code.
👉 For more insights on programmatic data labeling QA:
Awesome AI/ML resources: https://hubs.li/Q01-107f0
Social Media:
Subscribe to our YouTube channel: https://www.youtube.com/@kili-technol...
Follow us on LinkedIn: / kili-technology
Get the latest updates on Twitter: / kili_technology
Chapters:
0:00 - Writing Rules in Python
0:13 - Creating Custom Issues for Errors
0:25 - Triggering Plugin Upon Asset Submission
0:34 - Demonstration in User Interface
0:50 - Checking Plugin Run in Explore Page
1:03 - Correcting the Error and Marking Asset as Reviewed