In 2011, according to a Digital University Study, a staggering 1.8 zettabytes of information was created and duplicated. It is estimated that by 2020, we will generate 50 times that amount of information.  The impact of this growth has posed a peculiar burden upon the legal industry.

This is mainly due to the fact that in litigation or government investigations, attorneys have a legal duty to disclose all information related to a dispute. Today, that involves the production and review of a variety of electronic files across multiple servers, laptops or cloud databases.

For decades, linear review was the accepted standard within the legal industry.  That type of review required individual reviewers to manually review and hand code each document. Now conducting a strictly linear review is an outdated methodology.  It is simply not cost-effective for attorneys to review documents using the linear review workflow.

So, the next phase of development will be to adapt the review process to the capabilities that digital technology has to offer.  That means effectively integrating the technology into the traditional document review workflow.  I have employed the following workflow using predictive coding and other methods of assisted review:

• Legal team creates a sample set.
• The attorney review team codes the documents as responsive or not responsive.
• Using the sample set the analytic tool can analyze the collection to identify similar records and assign coding to the newly identified records based on sample set coded by humans.
• Attorneys approve or reject the newly identified documents.
• The tool uses that feedback to learn and identify similar records.
• This validation and retraining cycle repeats itself in several iterations.
• A statistical sampling of the not responsive documents is performed to validate that the records left are in fact not responsive.
• If the sample passes the statistical test for responsiveness, the review can be considered complete.

Ultimately, this can result in saving the costs associated with having humans review the entire data set.  Predictive coding can also be deployed for quality control purposes. For example, at the end of a document review, the tool can use a sample set of privileged documents to test against the entire document population.