Predictive Coding

January 22, 2016 Harvey Coblin

In my article on reducing eDiscovery costs in last month’s Sherpa newsletter, I made mention of an automated document review technique known as Predictive Coding. As discussed in the article, the primary expense involved in the eDiscovery process isn’t so much in the actual gathering of the relevant documents, but in the lengthy process of manually reviewing them to determine which are responsive, privileged or just plain irrelevant. Also known as technology-assisted review, predictive coding is intended to reduce eDiscovery costs by drastically reducing the time required for document review. But what exactly is it?
The idea behind predictive coding is to take a small sampling of all the potentially relevant case documents for manual attorney review. This is known as the “seed set.” Depending on the service provider, the seed set may be selected at random, or may be based on keyword search results or metadata, or a combination of the two. This subset is then categorized, or “tagged,” by attorneys or project managers. Once their relevancy of documents within the seed set has been determined, the predictive coding algorithm identifies patterns within them to analyze the broader case document collection, thereby dramatically speeding up the process of producing documents. The algorithm essentially “learns” from the manual categorization, and applies the same logic to the remaining case documents to determine which will be responsive.
Once the process is completed, senior attorneys perform a quality check against the results. If necessary, the process and be refined and repeated until the results are deemed acceptable.
Not only does this technique help meet court-imposed deadlines, but it’s also potentially helpful in helping attorneys assess the merits of their case. By identifying which documents are relevant much earlier in the process, attorneys can quickly decide if a case is worth fully pursuing or whether settlement might be a more suitable strategy.
But is predictive coding legally defensible? The jury is still out (no pun intended), but according to the Association of Corporate Counsel, “Both federal and state courts have generally approved the use of predictive coding technology as a means of fulfilling discovery obligations and maintaining costs proportional to the scale of the controversy.” As always, maintaining a clear audit trail is crucial.
To learn more about reducing eDiscovery costs in your organization, contact us (www.sherpasoftware.com/contact-us/) or visit www.sherpasoftware.com.

The post Predictive Coding appeared first on Sherpa Software.

Read more...

Previous Article
Sherpa’s Top 10 Must Reads!
Sherpa’s Top 10 Must Reads!

As 2016 kicks off, you’ve most likely heard “new year, new you” more times than you can count. New resoluti...

Next Article
How to Reduce eDiscovery Costs
How to Reduce eDiscovery Costs

There’s no doubt that litigation is expensive. When the added expense of producing and reviewing electronic...