Resources

SYNOPSIS

To support an international government investigation, the client needed to organize a 12.5 million document review that included multiple languages and a tight production deadline. 

CHALLENGES

SOLUTION

Using Brainspace, the leader in advanced eDiscovery data analytics, the population of 60,000 already reviewed documents were used to train a Continuous Multimodal Learning (CMML) model. CMML is an integrated set of features designed to support flexible interactive supervised learning workflows. The Predictive Ranks from the CMML model were then used to prioritize review ensuring the most relevant documents were reviewed first. Additionally, Brainspace’s patented Diverse Active Learning was used to build a small training round from a widely diverse data set to further optimize the Predictive Ranks.

RESULTS

By using Brainspace’s Continuous Multimodal Learning, the client was able to significantly reduce the review population from 1.8 Million to 280,000 docs, an 85% reduction in review volume. In addition, CMML produced a 3.7% richness level in a low richness data set. This was 270% higher than the less than 1% yielded by traditional keyword searching. With CMML, the client was able to find more relevant content faster, while reviewing far fewer documents. The savings on this case totaled approximately 19,000 hours of review time which is a cost savings of over $750,000 in attorney fees.