Core Concepts
Technology-Assisted Review (TAR)
Definition
A document review process that uses machine learning to predict which documents are relevant, rather than relying solely on linear human review.
Why it matters
It is the umbrella term for the workflows the rest of this glossary describes.
In practice
In technology-assisted review, people code a set of documents as responsive or not, and a machine learning model uses those decisions to score the rest of the collection. Reviewers then spend their time where the model says responsive documents are most likely to be, instead of reading everything in date or custodian order.
TAR does not remove people from the process. Lawyers decide what counts as responsive, review the documents the model prioritizes, and check the result with a random validation sample before anyone relies on it.
The two main families are TAR 1.0, which trains a model once and then classifies the collection, and TAR 2.0, which keeps learning throughout review. Newer review tools that use large language models to classify documents raise the same validation questions: how complete was the review, and how do you know.
Compare with
- Predictive Coding: An older name for the same idea, still common in court opinions and protocols.