Whether a legal firm is litigating an average case or a much larger dispute, however, the result is that legal teams are finding themselves sifting through thousands of gigabytes of data per case, with clients who aren’t always sure what they need, where it’s stored, or even if they’ve preserved all their data correctly in a defensible format. Successful eDiscovery demands a comprehensive document review process – but finding and tagging documents and data that are discoverable and relevant to the case is becoming increasingly more complex and time-consuming.
According to a 2015 study by the Rand Corporation, reviewing electronic documents makes up the largest percentage of eDiscovery production costs at 73% of all costs, collection consumes 8% of expenses, and the costs for processing are around 19% in typical cases.
To reduce the costs associated with the electronic review process, technology-assisted review (TAR) has gradually become standard practice in the eDiscovery process.
Key Benefits of Technology-Assisted Review (TAR)
Technology-assisted review (TAR) uses artificial intelligence and machine learning to analyze massive data sets, and then identify and tag potentially discoverable documents. It can provide statistics, categorization, and reporting data that is superior to traditional human review and requires less hours to produce.
If we consider that the average legal case in the US contains 6.5 million pages, 10 to 15 custodians, and 130 GB of data, any solution that simplifies electronic document review not only saves time but reduces the chances of human error or even missing documents entirely.
Key benefits of TAR include:
- All relevant documents are found and tagged. Before TAR software became a standard tool in the electronic document review process, teams of junior lawyers or paralegals would be required to manually review documents for eDiscovery. A 2005 study by Anne Kershaw compared an automated review system to a team of humans to determine the more efficient way to review documents. The software identified 95% of the relevant documents, while the team of humans only identified 51%.
- Legal firms are more protected against legal sanctions and fines. Companies in a legal dispute that fail to produce electronic data in an appropriate or timely manner risk paying millions of dollars in fines and sanctions, not to mention lost revenue, damaging the organization’s reputation and embarrassment.
- TAR increases access to courts. Because TAR reduces the time and costs associated with electronic document review, civil disputes that would have previously been settled to avoid lengthy legal battles are now more often taken to trial.
- TAR increases the likelihood of early case resolutions. Thanks to TAR, litigators can quickly evaluate the potential merits of a case, which will inform the likely value of the case, whether a settlement should be pursued, and support the legal team’s ability to formulate a comprehensive strategy early on in the dispute, mitigating a client’s risk and potential exposure and avoiding unnecessary litigation expense.
- Early access to information builds better cases. An early command of the facts gives litigators an important tactical advantage in structuring a case. For example, knowing whom to depose, which documents to seek from the opposing party, and the best affirmative defenses to pursue can mean the difference between winning and losing a dispute. Because TAR systems update in real-time, relevant documents will continue to be found as cases develop and trial strategies change.
How TAR Software Works
The TAR framework that TAR software is built on was developed in 2012 by an Electronic Discovery Reference Model (EDRM) team who mapped the necessary steps in a successful TAR process.
Broadly speaking, these steps are: determining key outcomes, setting protocols, educating reviewers, coding documents, predicting, testing and evaluating results, and then, ultimately, achieving the electronic review process’s goals.
Determining key outcomes of the entire review process could include the reduction and culling of non-relevant documents; prioritizing the most applicable documents; and quality control of the human reviewers involved in the process.
Next, review protocols are set, specifically around how document sets will be named and stored throughout the eDiscovery process.
The review protocol information must then be transferred to human reviewers prior to the start of the TAR Review.
Documents are then coded. During this step, a team of human reviewers code or tag a group of documents known as a “seed set,” which is uploaded into the TAR software platform in order to train the system. This process continues while the software learns which documents should be labeled with specific tagging codes, and to understand the boundaries of each category, such as the relevancy of a document. The quality of the process’s results depends heavily on the quality of the original seed set. If the seed set is flawed, the entire eDiscovery process will be flawed.
However, while TAR is not error-free, neither is human review. Technology-assisted review is more efficient and accurate than human review alone because it provides humans with a smaller, more relevant set of documents to review. And smaller sets of documents usually result in fewer errors.
At this point, the TAR system applies tags and classifies selected documents. These are validated by human reviewers using statistical sampling.
Finally, results are evaluated, and the review team determines if the TAR system has achieved its goals. If all review goals are met, the next phase of the review cycle can begin.
TAR and the Courts
According to Thomspon Reuters, courts today widely accept the use of TAR in the eDiscovery process, largely because TAR assists in reviewing vast amounts of documents, offers superior performance over other tools, is cost-effective, and is based on a transparent process.
The first case to uphold the use of TAR in eDiscovery was Federal Magistrate Judge Andrew Peck’s decision in Da Silva Moore v. Publicis Groupe in the Southern District of New York, 2012. Judge Peck determined that TAR is the best and most efficient tool in most cases.
The Limits of TAR
While there are many benefits to TAR, the biggest drawback is that it can only review text-rich documents like Word documents, emails, and PowerPoint slide decks. TAR can’t be used to review spreadsheets, images, videos, or blueprints. Without text information, there is nothing in the document for the software to identify.
A comprehensive eDiscovery strategy needs to include systems and processes that make it easy for humans to search, collect, and ultimately review content. While TAR is useful, a ‘black-box’ approach where humans have little insight into ESI doesn’t work.
Pagefreezer offers a range of options that, working together with TAR, can overcome limitations and assist humans with managing vast amounts of data that could be difficult for TAR to deal with – especially the kind of content that tends to dominate platforms like Facebook and Slack. Videos, GIFs, and emojis are all about context, and it helps if you have a human reviewer who can decide if that GIF or eggplant emoji is problematic.
Want to learn more? Download our case study, Facilitating eDiscovery of Workplace from Facebook Data, to see how Pagefreezer helped a leading global tech company streamline early case assessment of evidence in Workplace from Facebook.