Berger v. Graf Acquisition, LLC

What the court wrote

Verbatim from the court's document (public domain). Ellipses mark omitted text; nothing is paraphrased.

Fortunately, the defendants also propose a reasonable means to address their burden concerns: a technology assisted review (TAR). That is, the defendants are willing to review the universe of documents captured by the plaintiff’s broad search terms so long as they can employ a TAR protocol. But the plaintiff rejected this offer and insists that the defendants conduct a manual review instead. TAR, or predictive coding, leverages machine learning techniques with the guidance of attorneys to facilitate the document review process. The TAR system is trained with a representative seed set of documents selected and reviewed by attorneys for relevance. The system ingests this initial set of documents and undergoes iterative rounds of sampling and manual review, during which it learns and improves. Human reviewers remain involved in quality control, including by reviewing and running keyword searches over documents coded by the program as non-responsive. TAR promotes efficiency in the discovery process in several meaningful ways. ... Thus, the defendants may use TAR to reduce their discovery burden, so long as they are transparent with the plaintiff about their computer-assisted review process. Delaware counsel must remain closely involved in the review and sampling process for the TAR review.

Authorities cited in the passage: Moore v. Publicis Groupe, 287 F.R.D. 182 (S.D.N.Y. 2012) · Rio Tinto PLC v. Vale S.A., 306 F.R.D. 125 (S.D.N.Y. 2015) · Dynamo Holdings v. Commissioner of Internal Revenue, 143 T.C. 183 (2014) · Hyles v. New York City, 2016 WL 4077114 (S.D.N.Y. Aug. 1, 2016) · The Sedona Principles, Third Edition