E-Discovery: Use Predictive Tagging To Reduce Cost And Error

Tuesday, March 31, 2009 - 01:00

Editor: Please provide our readers with some background on yourself and Recommind.

Carpenter: Recommind sprang out of academia about nine years ago. A couple of professors from MIT, Brown University and University of California at Berkeley developed a technology that could discern themes and concepts within documents and relate them to similar themes and concepts in other forms of text. This was essentially a concept search tool, which was later applied to e-discovery and compliance. It really boils down to technology for conceptual search and automatic categorization of content irrespective of individual keywords.

I have been at Recommind for nearly two and half years. I am not only the general counsel, but also run marketing and get involved in a lot of customer engagements. I started my career at a law firm in the Bay Area doing transactional work for startups as well as litigation for small to mid-size companies. I also worked for several start ups in network security, information management and email spaces before coming to Recommind. Having experienced the life of a review attorney, I have a great appreciation for the rapid change of pace in the world of e-discovery - both from the process and technology perspective.

Editor: Is there concern that judicial standards for e-discovery might be overly demanding?

Carpenter: Not necessarily. If you look at a case like Victor Stanley [ Victor Stanley, Inc. v. Creative Pipe, Inc. , 2008 WL 2221841 (D. Md., May 29, 2008)], for example, Judge Grimm wasn't at all unreasonable. He basically said that it is not acceptable to reduce the universe of documents unilaterally just using keyword search without any sampling - and then doing a very poor job of that. He's absolutely right.

Another thing concerning judges who have addressed e-discovery - and what they're really trying to help the industry get better at - is that e-discovery in many cases may be far more expensive than the actual amount at issue in a given case. So, they are pushing both outside and corporate counsel to get a better handle on e-discovery costs at the outset.

Editor: Considering the current state of the economy, everyone is looking to reduce costs. Is e-discovery a good area on which to focus?

Carpenter: Very much so. Budgets for e-discovery were going steadily upward even before the economy started to stagnate. Companies recognize that e-discovery is a very good place to save money. While they have to spend a bit more money initially to bring core infrastructure in-house, they ultimately save significantly more money in not having to rely as much on outside law firms and service providers.

In-house counsel is focusing on how to effectively reduce the amount of data to be dealt with by doing data collection, preservation, processing and culling in-house. Our software helps at all four steps. If the corporate client buys our software and shifts the work in-house, it can save millions - if not tens of millions - of dollars.

Editor: Are you saying that the cost of investing in your software and doing the work in-house can save a significant amount of money?

Carpenter: Exactly. Spending a bit more to build more in-house infrastructure along the left side of the EDRM model is actually a way to ultimately reduce the budget. Less money is spent on outside litigation support providers because that function is brought in-house and the way you do things is more consistent and defensible because you have control over them. And, you don't have to reinvent the wheel each time you deal with different law firms with different approaches.

Editor: What is the EDRM model and what areas of the model do Recommind's products support?

Carpenter: EDRM stands for Electronic Discovery Reference Model. This is a vendor/constituent driven body tasked with setting up common e-discovery nomenclature and protocol standards. EDRM covers the spectrum of e-discovery, from information management through identification, preservation, review and analysis, production and ultimately presentation of electronically stored information at trial or as part of a mediation or arbitration. Recommind is the only vendor in this space that manages information for clients and outside counsel, addresses the clients' internal needs even before litigation ensues and offers full end-to-end information management and e-discovery solutions.

Editor: Is it beneficial from a cost standpoint to retain only one service provider to handle all e-discovery requirements?

Carpenter: Absolutely. They can go from many to only a primary and backup and negotiate significant discounting in return for that exclusivity.

Editor: It has been reported that the review phase of e-discovery is the most costly. What advances have you seen that help reduce those costs?

Carpenter: One is automated early case assessment. With it, data is indexed before anyone has reviewed a single document and the various concepts contained in the data are identified as well as the key people involved and the critical dates or topics. This is done not by a search using specific keywords but rather by using Predictive Tagging to search for themes and concepts. This approach provides a much better handle on the case from the very outset as well as a head start on organizing the content and data much more effectively.

Our approach channels the most pertinent documents to the most senior reviewers and those that are less relevant to the contract attorneys. Taking this a step further, some firms or clients might want to use our software to go through the 30 to 40 percent of data they know is spam or irrelevant, sample it in a defensible fashion and confirm that it need not be considered further. Or, they can use this technology to grab a small number of documents that are particularly relevant and plug them into our software to extract similar documents irrespective of individual keywords. Our software can actually go through and pre-code the entire data collection and essentially do an automated first pass review based on highly intelligent input from outside counsel.

Editor: Are the results you mentioned achieved by Predictive Tagging?

Carpenter: Yes, and Recommind is the only company in the industry that does it. Predictive Tagging is quite a bit more sophisticated than clustering, which is essentially a keyword based approach. Instead, Predictive Tagging groups documents based on their "aboutness" - their themes and concepts - rather than just individual keywords.

Editor: Does this sampling type review give a rough idea of what comprehensive e-discovery would cost, which may be a consideration in determining whether or not to settle a case?

Carpenter: Yes. This technology will allow you to do two things. First, it will do what we call DECA, Deep Early Case Assessment, that analyzes email strings to see what they are about and where the ones you really care about are located. Then, we also do an Early Risk Assessment, which considers how much the review process will cost and how much risk the case presents. Having a good initial understanding of how strong the case is and how much e-discovery is likely to cost allows you to come up with a range within which a case can and should be settled.

Editor: Have you seen corporate clients bring any phases of the EDRM model in-house? What phases are being brought in-house and why?

Carpenter: Yes, we've absolutely seen it. We've seen a huge interest in bringing the identification, preservation and collection processes in-house. Many companies had relied on outside service providers to handle this for them and it's a very expensive practice.

Editor: Would you say that the preliminary review is now so precise that having done that you're left with a fairly sophisticated stack of materials to analyze?

Carpenter: The process and workflow have become sufficiently defensible to permit companies to do this themselves if they so choose. Even contract reviewers are supplemented with software that is so highly sophisticated and very accurate that documents that shouldn't get missed aren't getting missed and documents that are false positives can be weeded out.

Editor: Do companies continue to struggle with certain areas of e-discovery?

Carpenter: Where a lot of companies struggle is information management - how much email you have, how much you save and how you save it, how much you delete if anything, and how you organize and find things. This is where it's critical to use e-discovery technology that understands and manages information, can find information in a sensible and accurate fashion, can automatically organize it and then provide the required information in a useable form. Recommind is ideally positioned because we've been dealing with data and information risk management for a decade; we're not a late entrant to the area.

Editor: What are some of the best practices in terms of data disposition you see being employed?

Carpenter: Folks who are really doing it well have a good handle on what data they need to keep, why they need to keep it, how they can find it and how they can organize it. This allows them to get rid of data they don't need.

Companies - like individuals - are basically pack rats. Having trouble finding things has always been an inconvenience, but lately it has actually become a real threat. The cost of e-discovery and compliance has gone up dramatically, in great measure because of having to deal with data that should have never been kept. The critical question is how to manage data in-house so that e-discovery and compliance can be as painless as possible while allowing employees to do their jobs. As I mentioned, our technology enables this to be done in the ordinary course of business - and in-house.

Editor: The ramifications of e-discovery are relevant not just to litigation but also to legislative and prosecutorial investigations at the state and federal levels. Do you have any thoughts in terms of what companies should be thinking about in terms of e-discovery?

Carpenter: Yes, indeed. In fact , I was at a conference two weeks ago where the head of enforcement for the SEC basically said that where you used to have perhaps a month or two to respond to an SEC information request, you now will need to respond within a day. Of course, this sent shock waves through the room. There certainly will be more litigation and also far greater regulatory scrutiny. Financial firms tend to be pretty well prepared because they've lived in this world for many years, but many others are going to be in a bit of trouble if they don't get prepared. More scrutiny, shorter timelines and greater sanctions are a recipe for disaster for folks who aren't ready.

Editor: If you had the opportunity to convey only one thought to a client about e-discovery, what would it be?

Carpenter: Don't be a spectator. It is not difficult or expensive to reduce e-discovery costs dramatically while improving the chances of keeping your company out of hot water in case of a law suit or an internal or external investigation. It takes an understanding of the problem and what it will take to meet your needs. Certainly, if you become educated and understand what you need to do, you'll be able to save money and actually become a hero internally.

Please email the interviewee at craig.carpenter@recommind.com with questions about this interview.