Predictive coding is revolutionizing eDiscovery document review by making it easier and more affordable than ever to identify relevant and privileged documents. Leveraging predictive coding for first pass review minimizes the costs associated with reviewing irrelevant documents. The savings are substantial because document review often approaches 80% of total eDiscovery costs.
Before predictive coding, keyword searching and exhaustive manual reviews were considered the “gold standard” for eDiscovery. However, these methods can be more error prone and require more time and effort than predictive coding. Studies show that predictive coding meets or exceeds the accuracy of human reviewers so there is no trade-off between quality and cost. As a result, law firms and corporate legal departments are adopting this technology and vastly improving their eDiscovery efficiency.
How Predictive Coding Works
Using complex algorithms, predictive coding software uses machine learning technology to analyze the manual coding decisions of human reviewers. The process begins by having someone with expert knowledge of the case manually review and code a sample set, or “seed set,” of documents. The software then analyzes these coding decisions and “learns” what criteria should be used for the remaining documents. It then “predicts” how the remaining documents should be coded and automatically applies those coding decisions. Once the process is complete, the results are quality checked and analyzed by a human reviewer for quality and completeness.