Review Fewer Documents with the Active Learning Coverage Queue
Active learning constantly updates its understanding of relevance in your project, based on what you code as “Responsive” and “Not Responsive.” The more examples it has to learn from, the more intelligent it becomes—and the faster you can power through review.
Now, you can help the model learn even faster. Introducing the coverage queue.
With this new queue, active learning serves up documents that most impact its understanding of relevance—typically the ones it’s not so sure about. It’s basically the engine asking, “Hey, what do you think of this document?"
By reviewing these documents first, you can help the model ramp up fast and, ultimately, get to the good stuff. As you work your way through the coverage queue, the engine gets smarter, and you get to review as few documents as possible.
Coverage queue is a great tool for helping active learning narrow in on what you need. For example, say you need to support a production workflow where you’re mandated to divide your documents into relevant and non-relevant groups. Coverage queue can help you get there quickly, without necessarily reviewing every single relevant document.