Machine-Assisted eDiscovery Processes For Cost Reduction, Defensibility And Quality Control

Monday, August 1, 2011 - 01:00

The Editor interviews Mary Mack , Enterprise Technology Counsel for ZyLAB.

Editor: Given the rapid development of eDiscovery and related technology, what steps is ZyLAB taking to assure that your clients are taking advantage of the most effective and economical cutting-edge technologies with respect to various aspects of Electronic Discovery Reference Model (EDRM) and the stages of ESI collection, preservation and production?

Mack: During ZyLAB's recent analyst tour, we were reminded that automated processes are highly sought after, yet perhaps not used to their fullest potential. That is likely due to lack of confidence on the part of users or the inability to explain just how the user reduced a terabyte of data (by 98 percent) to 20 GB for human review.

ZyLAB software provides the clarity, transparency and defensibility that legal professionals need in order to confidently apply automated eDiscovery tools. The ZyLAB system is truly end-to-end, managing and logging every action taken on data - from document creation, to legal or regulatory preservation, to review and production, and through to disposition of the document. At any stage, our self-service data sampling tool provides in-house and external resources with validation of the quality of the work performed. For example, we are seeing clients apply machine-assisted technology along with sampling to drive the disposition or expiration of documents. That allows a corporation to manage its information risk appropriately.

Editor: What distinguishes ZyLAB's approach to machine-assisted processes from other providers?

Mack: ZyLAB's model for indexing - the precursor to search and automatic coding - is better suited to eDiscovery than the algorithms that are under the hood on some similar products. We also support various other forms of machine-assisted processes to more fully complement the legal team. But I view the critical differentiator for ZyLAB's machine-assisted technology to be the process transparency and the ability for any team member to validate his or her human and machine-driven work and to make adjustments on the fly.

Editor: What do you mean when you say that some models of machine learning are better suited to eDiscovery?

Mack: ZyLAB is among those to offer machine-assisted processes that are based on faster mathematical models optimized for large, dynamic data populations. Once a document is processed, it is added to the index and is immediately available for coding, pattern recognition, concept or Boolean search. It is also ready for human review if none of the machine-assisted methods capture it. Other technology providers use different methods, many of which require a complete rebuild of the index before new material is available for machine or human review. During eDiscovery, rebuilding indexes for millions of documents can idle reviewers and delay productions. ZyLAB chose a better and more dynamic algorithm because our clients needed material to be made available very quickly, and they couldn't miss anything.

Editor: In what other ways can machine-assisted processes streamline eDiscovery responses and pragmatic information management within an organization?

Mack: Our clients are using ZyLAB technology in very strategic ways. For example, some are monitoring data by country or by language for compliance with the UK Bribery Act and FCPA. They can apply our pattern recognition tools to detect Social Security or account numbers, or to isolate gift giving, currency transactions and sentiment analysis, which identifies whether a message has a positive or negative message. Some clients use ZyLAB software to detect and automatically translate foreign language content, and others use it for proactive coding of information for data clean up, retention, legal hold, disposition or expiration. And, ZyLAB's Intelligent Redaction is helpful to inside and outside teams that need to protect personally identifiable information in advance of disclosures.

Editor: How can a corporate legal department or outside litigation team remain confident in their strategies and protocols when leaving so much to computerized execution? Is that what ZyLAB envisions with its Data Sampler tool?

Mack: Precisely. The ZyLAB Data Sampler provides [transparent] checks and balances so that practitioners don't feel like they are taking a leap of faith with machine processes. They can validate everything from search performance, to proportionality, to translation accuracy, to redaction accuracy, to human review quality. This technology does not replace lawyering - it enhances lawyering, allowing counsel to spend time (and billable dollars) constructing a trial and information strategy for optimal outcomes.

Jason Baron of the National Archives and Deb Logan of Gartner are building on the work of TREC to further legitimize machine-assisted review. The Sedona Conference and the EDRM have both put forth guidelines for using machine-assisted review. And Maura Grossman of Wachtell Lipton and Gordon Cormack of the University of Waterloo have published a wonderful article about the art of machine-assisted review.

I invite your readers to a webcast in which the topic of machine-assisted technology for eDiscovery will be discussed at length during an interview with Maura and Gordon:

Editor: Will eDiscovery technology ever be able to replace human review of large numbers of documents for relevance and privilege?

Mack: No, this technology won't replace lawyers, but it will reduce the number of lawyers doing a document review page by page at hourly rates. The work will be more fulfilling to the attorneys doing review, and more valuable to clients. The review attorneys will become integral to the work of data sampling and to the creation of the machine-assisted strategy, and the level of confidence required for the statistical analysis will be a legal decision. It is part of the advice given by outside counsel for risk evaluation by inside counsel.

For some cases, even one document inadvertently produced can impact a revenue stream or the company's reputation, and the best choice might be an eyes-on review for all such documents. In other cases, the risk of human eyes-on review is even greater than allowing a machine to make the review calls because "the walls have eyes." For such documents, it might be best to have the eyes-on review limited to a sample set viewed by only the most trusted of reviewers.

The revolutionary nature of powerful search, categorization and pattern recognition tools allows corporations unprecedented capacity to group documents by varying criteria and understand quickly what a division, department or employee has communicated about a particular subject. Machine-assisted review technologies are simple, additive tools that end-users in corporate and law firm environments can apply on the fly to influence their tactical and strategic steps. This is sometimes called early case assessment or early evidence assessment.

Editor: Given your evaluation of the present state of the art, to what extent do advances in eDiscovery technology help corporate legal departments and law firms become proactive in their data management and litigation readiness, particularly considering the proliferation of ESI through social and other new media?

Mack: With the cloud and social media creating another frontier unmanaged by the central information department, the same challenges around records management, legal hold, data destruction and harvesting the data for legal purposes suddenly exponentially increase. There will be no way to monitor without tools, so the tools legal counsel choose to use today should be able to collect from the cloud, and they should be able to parse and authenticate a social media post.

Editor: Are legal departments and law firms outsourcing more eDiscovery functions to technology partners? How does this relationship differ from the traditional "human review"-type outsourcing partnerships?

Mack: We find that our clients start by outsourcing services for an instant legal matter, and then go through the procurement and security due diligence to purchase the software in-house. We are seeing budgets freed up to bring eDiscovery software in-house, particularly for those who are not insured for their litigation, i.e., eDiscovery, costs. It only takes one medium-size matter to justify the purchase and installation of our software, for example. We use the same ZyLAB software we sell to our clients to provide our eDiscovery services, and we update our software with what we learn in our services business.

The powerful search and categorization capabilities engendered by machine-assisted review augment and enhance the strategic nature of legal collaboration, and they reduce the costs related to repetitive actions.

Editor: Will technological advances that can significantly affect cost cause corporate legal departments to handle more eDiscovery in-house? How does the growth of machine-assisted processes impact staff workloads and billables?

Mack: Even with the growth of machine-assisted eDiscovery processes, legal departments will handle some eDiscovery in-house, but not all. Generally, a legal department will not want to invest in resources necessary to be prepared for the outlier suits, like huge class actions. This is better handled by the law firm or service provider, which have the infrastructure and personnel to handle it. Some corporate legal departments will purchase the software and have outside counsel or provider personnel operate it, and others will train their own IT personnel or paralegals to run the software. It is safe to say that more corporate legal departments want the benefit of the information system knowledge gained in the identification, preservation and collection stages, the review calls for privilege and the coding and production status of their documents. Having the document population with work product is a valuable corporate asset. It can be used for good faith productions without the costs of reviewing the same document again and again. (Increasingly, law firms are being asked to work with the client's choice of technology for that very purpose.) Law firms who become expert in their client's choice of technology become the go-to firm for cases with a lot of eDiscovery.

In a law firm setting, outside counsel love ZyLAB's self-service sampling tool. This gives them the ability to create court-consumable documentation for their due diligence in making a "reasonable inquiry." It makes it more comfortable for an attorney to sign an FRCP 26(g) certification without a page-by-page review or processing all potential custodians that have received a legal hold notice.

In the end, properly documented, machine-assisted review will stop eDiscovery from being the tail that wags the dog in litigation and settlement.

Please email the interviewee at with questions about this interview.