23 Oct 2012

What’s in a Name? Understanding the Terminology Surrounding E-Discovery’s Predictive Phenomenon

Posted in: Tuesday Trends No Comments

By: Andrew Bartholomew

It is fair to say that the e-discovery market has been inundated by everything “predictive” of late. For e-discovery practitioners this is great news. The application of machine intelligence to make sense out of exponentially expanding volumes of electronic information, long utilized in other industries, is already transforming how legal teams go about identifying relevant information.

While we are starting to get a general idea of how the technology works, there doesn’t seem to be consensus on what we’re supposed to call it. I interviewed prominent e-discovery attorney Ralph Losey a few months back and he expressed frustration with the lack of common terminology around the new technology. This all begs the question: do names really matter? They do when the name itself conveys the perceived utility of the product.

Predictive coding appears to be the label “du jour,” but there are some problems with the universal use of that term. For starters, predictive coding is a vendor-ascribed product name and, therefore, inherently reflects that vendor’s interpretation of how the technology works and how it should be applied. For this reason, some choose to use more descriptive, neutral labels, such as computer-assisted review or technology-assisted review. However, these labels have their own shortcomings. Technology has long been utilized for document review. A basic keyword search tool, for instance, is a form of technology that many legal teams already utilize to hone in on relevant documents. But it hardly represents the truly advanced machine intelligence that is now available on the market.   

Another problem with the terms described above is that they have all come to describe the application of technology solely during e-discovery review. This is understandable. The first widely accepted application for predictive technologies has been aimed at helping legal teams expedite the traditionally grueling, expensive review process. Indeed, most vendors that have introduced their own predictive products have conformed to this paradigm. But the application of predictive technologies is moving beyond the review phase. E-Discovery analyst Christine Taylor of the Taneja Group recently wrote, “given a good sample set and good science, machine learning offers a lot of benefits throughout the e-discovery process.”

To that end, Exterro last week launched Fusion Predictive Intelligence™, a new predictive technology that applies machine learning for the identification and categorization of electronically stored information (ESI) across all e-discovery stages, from pre-collection through review. “Exterro’s brand of predictive technologies works as early as the identification stage,” wrote Taylor. “Once a model is trained, the team directs the tool to potentially relevant documents and emails for fast identification, analysis and classification.”

The purpose here is not to simply extol Exterro and its new technology offering. Rather, it is to draw attention to the fact that the traditional labels that get thrown around aren’t necessarily appropriate for all technologies. Already, some reporters and analysts have taken to describing Exterro’s Fusion Predictive Intelligence as not another predictive coding product. Commenting on Exterro’s launch last week, eDJ Group analyst Barry Murphy used the words “unique” and “refreshing.” Time will tell how fast the market follows suit and moves beyond the “predictive coding” mold.

To learn more about how predictive technologies are being applied across the EDRM, tune into Exterro’s upcoming webcast, “Leveraging Predictive Technologies across the EDRM.”

No Responses to “What’s in a Name? Understanding the Terminology Surrounding E-Discovery’s Predictive Phenomenon”

Leave a Reply