In an assisted review, what minimum coherence score does the system bump to?

In an assisted review process, the coherence score represents the system's confidence in its ability to classify documents accurately based on prior review data and machine learning algorithms. The system typically establishes a threshold that determines which documents are deemed suitable for review by legal professionals.

The minimum coherence score to which the system bumps is set to 70. This level is chosen to ensure that the documents being presented for review are of a high confidence level, reducing the risk of misclassification and improving review efficiency. Maintaining this threshold allows review teams to focus their efforts on documents that are more likely to be relevant to the case at hand, while also optimizing the overall workflow of the review process.

A score lower than this might result in unnecessary reviews of documents that the system is unsure about, potentially leading to inefficiencies and errors in judgment. Therefore, the choice of 70 strikes a balance between reliability and practicality in the assisted review workflow.