Predictive Review is the practical application of machine learning technology to e-discovery. Predictive Review combines attorney review with advanced machine learning to substantially improve the efficiency and accuracy of identifying relevant documents in litigation and regulatory investigations.

Predictive Review involves the attorney review of a sample of documents that is used to train the Servient learning system. Servient performs statistical analysis of various features contained in the document set and evaluates the attorney decisions on the reviewed documents. Servient then evaluates all of the non-reviewed documents, assigning each document a probability of relevance.

The more documents the legal team reviews, the more Servient learns and the more accurate the automated document decisions become.

Servient refers to its technology as "active" learning because Servient actively selects the documents that should be reviewed by the legal team to achieve the best accuracy with the least amount of manual review. Servient's learning technology is tightly integrated into its full-featured e-discovery platform; therefore the legal team does not have to change their normal process. Instead, the lawyer reviews documents in a familiar interface and workflow while the advanced technology performs the statistical analysis, controls the document assignments and validates the reliability of the process.

Predictive Review unlocks powerful new options for the legal team to improve the quality of the e-discovery process and achieve substantial cost savings. Servient is truly "game changing" technology that fundamentally alters the economics of e-discovery.