Technology-assisted review, also known as “predictive coding,” has become a permanent part of the e-discovery toolkit. Methodologies that automated parts or all of the review process started to gain traction a few years ago with forward-thinking legal teams seeking to transform the economics of e-discovery for their large-scale matters (Conduent introduced its own automated document classification tool, CategoriX, more than two years ago). Its adoption was recently propelled forward, though, by a spate of court cases, most notably Magistrate Judge Andrew Peck’s Da Silva Moore, et al. v. Publicis Groupe, et al.
But has technology-assisted review replaced predecessor technologies and methodologies like keyword search? In my view, the answer is a resounding no – despite perceptions that technology-assisted review and keyword search are mutually exclusive. (This was the subject of a recent article I authored in Metropolitan Corporate Counsel).
It has been our experience that both cost-efficiency and knowledge acquisition can be maximized in e-discovery by combining complementary tools and methodologies. Utilizing keyword search strategically in conjunction with technology-assisted review, for instance, allows users to benefits from the strengths of both techniques while minimizing the weaknesses. Eliminating keyword search wholesale in favor of technology-assisted review unnecessarily forfeits the many benefits keywords have to offer.
In Da Silva, the court acknowledged that keywords combined with predictive coding can be instructive. In that case, for example, the protocol stipulated that keywords would be used to augment the training population and ensure that a sufficient number of documents were identified for key topics of interest in order to ensure that the system was robustly trained in those areas.
In the article I also note these additional examples of how keyword searching can help optimize a technology-assisted review process:
- During culling, utilizing thoughtfully developed and thoroughly tested keyword searches can help save significant downstream hosting costs by eliminating patently irrelevant material. It will simultaneously help to ensure that the technology-assisted review tool has access to a population rich in relevant data, facilitating more effective and efficient training.
- If there are extremely rare but important topics in the document collection, keyword searching can help capture those documents in a way that random sampling alone will not, enhancing the system’s ability to target and retrieve key content.
- Keywords offer ways to strategically capitalize on the explicit knowledge and insights of the attorneys working on a case. The statistical algorithms that drive technology-assisted review are designed to discover latent patterns of relevance in the review population. Keywords offer ways to incorporate the unique and complementary perspectives of humans.
E-discovery poses increasingly complex challenges every day. To cope with the ever-growing burden of document review, we need all of the tools at our disposal – including both keyword search and technology-assisted review.
About the Author