A funny thing happened on the way to the webinar. On Wed Dec. 14th, Advanced Discovery is presenting a live webinar entitled “New Developments in Analytics”. (link and full information at the end of this article) But while preparing the slide deck and speaking with several of our internal experts on the AD Consulting Team (special thanks to Susan Stone, Julia Byerson and Todd Mansbridge for all their feedback) as well as several clients about the topic, I found that we had a surprising lack of agreement on some of the key terms.
I had always viewed TAR (Technology Assisted Review) as the granddaddy of this discussion because many years ago I felt that TAR essentially meant keyword searching. I also felt it then evolved into Predictive Coding and later in the game the phrase ”analytics” was grabbed from big data people to refer to some data analytics tools.
So my world view of TAR looked something like this:
• Email threading
• Near Duplicate detection • Language detection
• Keyword expansion
• Conceptual clustering
• Predictive Coding
And a recent article by another vendor expressed the view that there are three classes of analytics – structured, conceptual and predictive, with predictive including TAR.
Finally, this graphic from Relativity shows that their world view of RAR (Relativity Assisted Review) appears to be one all-encompassing definition.
But other people were looking at these terms from a different perspective. One of Solutions consultants elaborated that:
Conceptual indexing is an internal (non-client facing) analytics tool.
Predictive coding is a class of workflows that can sit on top of different internal analytics tools.
RAR is a product (or a feature of a product) that combines both the internal analytics tool of conceptual indexing with a repeatable, defined predictive coding workflow.
Another of our experts expressed it much more simply:
Predictive coding is a process not a product or service.
And of course, I had to add to the confusion by asking where CAL (Continuous Active Learning) fit into this hierarchy. One of our senior analytics gurus responded to that query with:
In my opinion CAL is a workflow. It selects seed documents based on categorization which is live rather than passive which requires you to submit after you have reviewed a set number.
Finally, when I ran all this by Matthew Verga, the Advanced Discovery VP of Marketing Content, he fell clearly in the tools vs workflow camp, saying:
TAR is process that uses analytic tools to amplify human decision-making. Relativity Assisted Review is a form of TAR [and] is powered by categorization, an analytic tool. Neither Analytics nor TAR contain the other as a subset, they are different categories of things
And indeed it is this workflow paradigm that is much more prevalent today, as we will discuss in the webinar. But what I’m interested in is hearing what YOU think. Analytics, TAR, Predictive Coding …. how do YOU define these terms and how do you use them in your work with ESI?
Drop me a note at firstname.lastname@example.org and join us on Wednesday at 1pm EST for an hour-long nuts and bolts discussion about how to use technology in your eDiscovery practice. I’ll be joined by Anne Bentley McCray from McGuireWoods LLP, an attorney experienced in working with ESI to discuss these concepts and others.
The registration page can be found at: https://attendee.gotowebinar.com/register/1528164311149570051