After early concerns about the defending the results of the technology and whether courts would accept its use, Technology Assisted Review (“TAR”) has now entered the spotlight as an alternative to more traditional forms of document review. These technologies, commonly referred to as predictive coding, continue to win over both clients and counsel, who have achieved significant efficiencies, cost savings and improved results over more traditional review options, including keyword searching and manual review. The strength of TAR processes is that they harness the judgments of the most knowledgeable human reviewers – the subject matter and case team experts – by having those experts train the software by coding relatively small sample sets of documents. The platform applies these judgment calls to the full data set, which often includes millions of documents, in an iterative process of multiple training rounds. Then the experts and the project managers review the results of the coding process and engage in thorough quality control efforts to ensure that the results are appropriate. When executed by experienced attorneys and review managers, the TAR process yields a high level of accuracy and consistency with only a fraction of the documents requiring manual review. In our “Data Law Trends & Developments: E-Discovery, Privacy, Cybersecurity & Information Governance” publication, we examine these topics and address some of the common components of TAR workflows, including the selection of “seed set” documents, the management of training rounds and how to effectively incorporate non-TAR quality control metrics. We also discuss how comparisons between manual review – once considered the “gold” standard – and TAR have shown quantitatively that TAR is at least, if not more, effective and often far more cost-efficient.
Bloomberg BNA’s Digital Data and e-Evidence publication also features our discussion: Technology Assisted Review Goes Mainstream.