Intro to Predictive Coding: Overview & Interpretation of Terminology June 2014 | Page 21

Glossary
Active learning – a form of supervised machine
learning that presents for review or human
categorization the documents with the highest
current uncertainty, those documents that will be
most informative about how to update the learning
process.
Bayesian categorizer—an information retrieval tool
that computes the probability that a document is a
member of a category from the probability that
each word is indicative of each category. These
estimates are derived from example documents.
Uses the probability of each word given each
category to compute the probability of each
category given each word. Also called a naïve
Bayesian Categorizer.
CAR – Computer assisted review. Any of a number
of technologies that use computers to facilitate
the review of documents for discovery. See TAR.