Technology assisted review (TAR) is one of the eDiscovery industry’s top buzzwords today, and there are a variety of litigation support software options to go with it. For example, Catalyst and IPRO both offer predictive coding technology. We were one of the first electronic discovery firms in the industry to work with Equivio’s Zoom tool inside the Catalyst platform.
How does predictive coding work?
Attorneys perform a first pass review of a statistically valid sample of the data population to “train” the predictive coding program to recognize and code documents accurately. The litigation support software then applies this knowledge to the entire universe of documents to identify, prioritize and weigh documents based on similarity.
The Association of Corporate Counsel recently published their top 10 concepts to understand about predictive coding. Here are some of our favorite points:
- Technology assisted review (TAR) is more than predictive coding. Although TAR and predictive coding are often used interchangeably, predictive coding is just one part of TAR, which also includes tools that identify near-duplicate documents and latest-in-thread emails.
- In predictive coding, a seed set is a sample of all of the documents in the population, and it will be reviewed by attorneys. The determinations made on the seed set are used to teach the predictive coding machine, which finds patterns of relevance throughout the entire document set.
- Predictive coding uses iterative learning, meaning the tool continues to learn more about the data as the process continues and can make on-the-fly adjustments.
- It’s defensible. So far both federal and state courts have generally approved the use of predictive coding technology as a means of fulfilling discovery and maintaining costs proportional to the scale of the case.
Is predictive coding only for large cases?
No. In fact, we recently used the method on a relatively small data set of 50,000 documents (after deduplication and date filtering). In this case, based on the richness of the data set and the associated feedback from the predictive engine, the TAR platform determined the amount of documents requiring a “second pass” review and eliminated the rest from the client’s review process.
Even in this relatively small case, the litigation support tool saved our client about 640 hours of manual review, which equates to at least $176,000 in savings if you assume $275 an hour for review.
73% of eDiscovery costs are spent on attorney fees while only 19% comes from technology, according to a February 2013 study by the RAND Corporation’s Institute for Civil Justice. The technology assisted review tools that we offer, like Equivio Zoom, Catalyst’s Insight Predict and IPRO’s Eclipse, are saving our clients significant amounts of time and money.
At DSi, we work with these and other litigation support tools and techniques to customize solutions for each client. As document-by-document review becomes physically and financially unfeasible, there will be an even greater reliance on technology to analyze data. We are committed to being ahead of the technology curve, and our predictive coding tools are just one example of this commitment.
Click here to download a detailed study about the case mentioned above and learn more about how DSi saved one client more than $176,000 on a small case by using predictive coding.