Using technology assisted review in the right cases and in the right way
Technology assisted review also referred to as predictive coding or computer-assisted review has evolved into a viable method to gather, review and produce relevant documents in litigation.
November 14, 2013 at 03:00 AM
7 minute read
The original version of this story was published on Law.com
This is the first of a three-part series concerning technology assisted review of documents in civil litigation. In the first part, we assess various methods of such review and discuss when it should be considered for use in civil litigation. The second and third parts will focus on how to make technology assisted review work for your company in litigation while achieving cost savings.
Your company is engaged in a large civil action, and discovery has begun. The discovery requests you have received from opposing counsel seek voluminous electronic data, and the idea of using key word searches and manual attorney review strikes you as both unwieldy and costly. As recently as five years ago, you would not have had any attractive alternatives. Now, however, technology assisted review (TAR) — also referred to as predictive coding or computer-assisted review — has evolved into a viable method to gather, review and produce relevant documents in litigation. The pertinent questions become whether you should use TAR, and how best to fit its use within your company's broader litigation strategy.
Given its relative newness, not all attorneys understand what TAR is or how it works. In a nutshell, TAR involves taking a subset of documents reviewed by attorneys and marked for responsiveness and then, using specialized software, applying the review decisions made for that subset to all remaining un-reviewed documents. In supervised learning TAR, a type of predictive coding, the software simply looks for documents that have the same characteristics as the relevant documents in the subset. In active learning TAR, also a predictive coding method, you start with an initial “seed set” chosen by attorneys and fed into the program, which returns samples of potentially responsive documents and asks an attorney to decide whether they are, in fact, responsive. This method is more iterative and interactive. Finally, in knowledge engineering TAR, software replicates how an attorney thinks about complex problems, creating a decision tree (artificial intelligence) that the computer uses to determine whether a document is responsive. Recent research has shown that TAR produces results that are at least comparable to manual review, with active learning and knowledge engineering TAR demonstrating 10 percent greater accuracy.
Companies and their counsel need to consider the nature of the case when deciding whether to use TAR. For example, small matters with low financial stakes, few informational custodians, or limited amounts of electronic data are not good candidates for TAR. Cases in which responsive electronic data contains images, abundant numerical data, short text messages or complex coding schemes are also not ideally suited for TAR due to the nature of that information. On the other end of the spectrum, those dealing with cases involving high financial stakes, large volumes of mostly textual electronic data (such as e-mail), and short discovery periods will find TAR particularly helpful. In these cases, proportionality concerns should be foremost, as courts are becoming more willing to reduce the scope of discovery requests if searching for and producing responsive documents is more costly than the value of the case itself.
If you believe TAR would help make discovery in a particular case more efficient and manageable, you should ensure that your company's attorneys are knowledgeable about the technology and engaged throughout the e-discovery process. Thoughtful advice at the subset or seed set stage of review, and continued involvement in the application of TAR, will improve both the accuracy of the results and the likelihood that the court will approve them. Working closely and effectively with opposing counsel will also help ensure that TAR withstands judicial scrutiny.
As the technology advances and more lawyers become familiar with the advantages TAR offers, an ever greater number of parties are likely to use its methods in electronic discovery. The final two parts of this series will provide practical advice on managing the use of TAR vis-à-vis opposing counsel and achieving cost efficiency.
This is the first of a three-part series concerning technology assisted review of documents in civil litigation. In the first part, we assess various methods of such review and discuss when it should be considered for use in civil litigation. The second and third parts will focus on how to make technology assisted review work for your company in litigation while achieving cost savings.
Your company is engaged in a large civil action, and discovery has begun. The discovery requests you have received from opposing counsel seek voluminous electronic data, and the idea of using key word searches and manual attorney review strikes you as both unwieldy and costly. As recently as five years ago, you would not have had any attractive alternatives. Now, however, technology assisted review (TAR) — also referred to as predictive coding or computer-assisted review — has evolved into a viable method to gather, review and produce relevant documents in litigation. The pertinent questions become whether you should use TAR, and how best to fit its use within your company's broader litigation strategy.
Given its relative newness, not all attorneys understand what TAR is or how it works. In a nutshell, TAR involves taking a subset of documents reviewed by attorneys and marked for responsiveness and then, using specialized software, applying the review decisions made for that subset to all remaining un-reviewed documents. In supervised learning TAR, a type of predictive coding, the software simply looks for documents that have the same characteristics as the relevant documents in the subset. In active learning TAR, also a predictive coding method, you start with an initial “seed set” chosen by attorneys and fed into the program, which returns samples of potentially responsive documents and asks an attorney to decide whether they are, in fact, responsive. This method is more iterative and interactive. Finally, in knowledge engineering TAR, software replicates how an attorney thinks about complex problems, creating a decision tree (artificial intelligence) that the computer uses to determine whether a document is responsive. Recent research has shown that TAR produces results that are at least comparable to manual review, with active learning and knowledge engineering TAR demonstrating 10 percent greater accuracy.
Companies and their counsel need to consider the nature of the case when deciding whether to use TAR. For example, small matters with low financial stakes, few informational custodians, or limited amounts of electronic data are not good candidates for TAR. Cases in which responsive electronic data contains images, abundant numerical data, short text messages or complex coding schemes are also not ideally suited for TAR due to the nature of that information. On the other end of the spectrum, those dealing with cases involving high financial stakes, large volumes of mostly textual electronic data (such as e-mail), and short discovery periods will find TAR particularly helpful. In these cases, proportionality concerns should be foremost, as courts are becoming more willing to reduce the scope of discovery requests if searching for and producing responsive documents is more costly than the value of the case itself.
If you believe TAR would help make discovery in a particular case more efficient and manageable, you should ensure that your company's attorneys are knowledgeable about the technology and engaged throughout the e-discovery process. Thoughtful advice at the subset or seed set stage of review, and continued involvement in the application of TAR, will improve both the accuracy of the results and the likelihood that the court will approve them. Working closely and effectively with opposing counsel will also help ensure that TAR withstands judicial scrutiny.
As the technology advances and more lawyers become familiar with the advantages TAR offers, an ever greater number of parties are likely to use its methods in electronic discovery. The final two parts of this series will provide practical advice on managing the use of TAR vis-à-vis opposing counsel and achieving cost efficiency.
This content has been archived. It is available through our partners, LexisNexis® and Bloomberg Law.
To view this content, please continue to their sites.
Not a Lexis Subscriber?
Subscribe Now
Not a Bloomberg Law Subscriber?
Subscribe Now
NOT FOR REPRINT
© 2025 ALM Global, LLC, All Rights Reserved. Request academic re-use from www.copyright.com. All other uses, submit a request to [email protected]. For more information visit Asset & Logo Licensing.
You Might Like
View AllInternal Whistleblowing Surged Globally in 2024, So Why Were US Numbers Flat?
6 minute readInside Track: AI Is Sure to Fray Big Law's Devotion to Billable Hour
Trending Stories
- 1New York-Based Skadden Team Joins White & Case Group in Mexico City for Citigroup Demerger
- 2No Two Wildfires Alike: Lawyers Take Different Legal Strategies in California
- 3Poop-Themed Dog Toy OK as Parody, but Still Tarnished Jack Daniel’s Brand, Court Says
- 4Meet the New President of NY's Association of Trial Court Jurists
- 5Lawyers' Phones Are Ringing: What Should Employers Do If ICE Raids Their Business?
Who Got The Work
J. Brugh Lower of Gibbons has entered an appearance for industrial equipment supplier Devco Corporation in a pending trademark infringement lawsuit. The suit, accusing the defendant of selling knock-off Graco products, was filed Dec. 18 in New Jersey District Court by Rivkin Radler on behalf of Graco Inc. and Graco Minnesota. The case, assigned to U.S. District Judge Zahid N. Quraishi, is 3:24-cv-11294, Graco Inc. et al v. Devco Corporation.
Who Got The Work
Rebecca Maller-Stein and Kent A. Yalowitz of Arnold & Porter Kaye Scholer have entered their appearances for Hanaco Venture Capital and its executives, Lior Prosor and David Frankel, in a pending securities lawsuit. The action, filed on Dec. 24 in New York Southern District Court by Zell, Aron & Co. on behalf of Goldeneye Advisors, accuses the defendants of negligently and fraudulently managing the plaintiff's $1 million investment. The case, assigned to U.S. District Judge Vernon S. Broderick, is 1:24-cv-09918, Goldeneye Advisors, LLC v. Hanaco Venture Capital, Ltd. et al.
Who Got The Work
Attorneys from A&O Shearman has stepped in as defense counsel for Toronto-Dominion Bank and other defendants in a pending securities class action. The suit, filed Dec. 11 in New York Southern District Court by Bleichmar Fonti & Auld, accuses the defendants of concealing the bank's 'pervasive' deficiencies in regards to its compliance with the Bank Secrecy Act and the quality of its anti-money laundering controls. The case, assigned to U.S. District Judge Arun Subramanian, is 1:24-cv-09445, Gonzalez v. The Toronto-Dominion Bank et al.
Who Got The Work
Crown Castle International, a Pennsylvania company providing shared communications infrastructure, has turned to Luke D. Wolf of Gordon Rees Scully Mansukhani to fend off a pending breach-of-contract lawsuit. The court action, filed Nov. 25 in Michigan Eastern District Court by Hooper Hathaway PC on behalf of The Town Residences LLC, accuses Crown Castle of failing to transfer approximately $30,000 in utility payments from T-Mobile in breach of a roof-top lease and assignment agreement. The case, assigned to U.S. District Judge Susan K. Declercq, is 2:24-cv-13131, The Town Residences LLC v. T-Mobile US, Inc. et al.
Who Got The Work
Wilfred P. Coronato and Daniel M. Schwartz of McCarter & English have stepped in as defense counsel to Electrolux Home Products Inc. in a pending product liability lawsuit. The court action, filed Nov. 26 in New York Eastern District Court by Poulos Lopiccolo PC and Nagel Rice LLP on behalf of David Stern, alleges that the defendant's refrigerators’ drawers and shelving repeatedly break and fall apart within months after purchase. The case, assigned to U.S. District Judge Joan M. Azrack, is 2:24-cv-08204, Stern v. Electrolux Home Products, Inc.
Featured Firms
Law Offices of Gary Martin Hays & Associates, P.C.
(470) 294-1674
Law Offices of Mark E. Salomone
(857) 444-6468
Smith & Hassler
(713) 739-1250