The increasing cost of eDiscovery remains a significant challenge for litigants, particularly corporate defendants. But there is reason for hope. A recent study conducted by TREC, an ongoing workshop sponsored by the National Institute of Standards and Technology, adds to a growing body of literature demonstrating that the use of “concept” and “predictive coding” software can substantially reduce eDiscovery costs.

The 2010 Text Retrieval Conference challenged legal document review vendors to employ their advanced searching software against publicly available documents from the Enron Bankruptcy case. One of those vendors, Equivio Inc., did so and reported that use of its predictive coding software could potentially allow a legal team to retrieve nearly 80 percent of the documents responsive to discovery requests by reviewing less than 4 percent of the document pool—a substantial reduction in attorney review time as compared to manual document review. These results reinforce the finding of a 2009 TREC study of computer-assisted review software, which also demonstrated superior results for an advanced searching technology known as H5 as compared to manual attorney review.

The process of predictive coding entails a number of steps designed to retrieve a set of relevant documents for second-level attorney review. The software is “trained” to predict relevant keywords and phrases through an iterative process. First, lawyers develop search criteria, which the software then uses to search, cull, and cluster documents it determines to be responsive to the search query.

“Concept searching” tools expand the effectiveness of the search criteria by identifying “like” documents using mathematical algorithms. Lawyers then review the results, analyze the effectiveness of the concept searching, and adjust the search criteria to include or omit various types of documents returned. Over time, the software “learns” how to recognize responsive documents and eliminate non-responsive documents from the review set. This process can lead to dramatically lower costs and improved consistency in document searching.

Although promising, the TREC results should be viewed somewhat cautiously. Using advanced searching technologies to achieve cost savings in a particular litigation is not, as some may believe, simply a matter of “pressing a button.” Working effectively with these tools is a sophisticated and learned skill. Among other things, attorneys are necessary to “train” the concept searching engine. Moreover, given the investment of time needed to train the concept searching engine, the technology is most efficient when applied to large document reviews that extend over a course of several months and may not make sense for smaller litigations.

Still, the results of the 2010 TREC study are welcome news to litigants with high eDiscovery bills. For larger document reviews where computer-assisted review is appropriate, litigants should consider working with a law firm or vendor familiar with concept searching and predictive coding tools. For questions or more information about these tools or minimizing costs, contact Philip N. Yannella at 215.864.8180 or, or Carl G. Roberts at 215.864.8120 or


Copyright © 2011 by Ballard Spahr LLP.
(No claim to original U.S. government material.)


All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, including electronic, mechanical, photocopying, recording, or otherwise, without prior written permission of the author and publisher.

This alert is a periodic publication of Ballard Spahr LLP and is intended to notify recipients of new developments in the law. It should not be construed as legal advice or legal opinion on any specific facts or circumstances. The contents are intended for general informational purposes only, and you are urged to consult your own attorney concerning your situation and specific legal questions you have.