Core Concepts
Continuous Active Learning (CAL)
Definition
A TAR approach where the model is continuously retrained on reviewers' coding decisions, constantly reprioritizing the highest-value documents to review next.
Why it matters
It is the dominant modern workflow and changes how you decide when review is complete.
In practice
Continuous active learning is a protocol where the model is retrained as reviewers code documents, and the next documents shown to reviewers are the ones the model currently ranks as most likely responsive.
The term was popularized by Gordon Cormack and Maura Grossman, whose research compared CAL with other TAR protocols and found it generally reached high recall with less review effort.
Because CAL keeps finding the likeliest documents first, the share of responsive documents in each new batch eventually drops. That decline is the usual signal to consider stopping, followed by validation.
Compare with
- TAR 2.0: TAR 2.0 is the workflow label; CAL is the learning protocol behind most of those workflows.