Catalyst Improves Algorithm for Better Machine Learning, Faster Review
Global News

Catalyst logoCatalyst, the leader in fast and scalable e-discovery platforms, today debuted a significantly improved version of the algorithm at the heart of Predict, the award-winning technology assisted review tool for the Insight and Insight Enterprise platforms.

The new Predict algorithm has been re-engineered to further reduce the total cost of discovery—by significantly improving speed and accuracy.

As the first commercial review product to use an advanced continuous active learning protocol—one of the fastest and most-effective TAR methods—Predict was already named the “Best New Product of the Year” by the LTN Innovation Awards.

“The best just got better,” said Catalyst founder and CEO John Tredennick. “This major algorithm enhancement helps further reduce the number of documents to review while increasing the speed by up to 40 percent. By controlling our own technology, we can constantly improve on what was already the best TAR platform available.”

Insight Predict is a second-generation TAR engine designed to reduce review costs and time for outbound productions, deposition prep, early case assessment and investigations. Predict uses a form of artificial intelligence (AI) to rank documents by their likely relevance, so that reviewers see relevant documents first. By eliminating review of irrelevant documents, Predict can reduce the cost of a project by 80 percent or more.

Legal IT Today #18

Today’s update includes enhanced parsing of document text to expand the feature set of the supervised machine learning Predict algorithm to include multi-word bigrams and trigrams. These new features will further improve Predict’s ability to differentiate relevant documents and elevate them in the ranking for early, prioritized review.

"These improvements will significantly improve batch richness in the early stages of the Predict review, delivering better documents even sooner. Ultimately, this enhanced algorithm will improve efficiency by reducing the number of documents reviewed and the associated cost by 20 to 40 percent,” said Thomas Gricks, managing director of Professional Services. “The savings for 250,000 documents could range from $50,000 to $100,000.”

The new algorithm is already integrated into the Catalyst platform and in use on multiple new cases.


Copyright © 2016 Legal IT Professionals. All Rights Reserved.

Daily News Alert

All the legaltech headlines in your mailbox, sign-up now! 

Media Partnerships

We offer organizers of legal IT seminars, events and conferences a unique marketing and promotion opportunity. Legal IT Professionals has been selected official media partner for many events.

A muchbeta site