In October of 2018, the U.S. District Court for the District of Columbia (D.C.) ruled in favor of government agencies making a “good faith effort” when responding to FOIA requesters, beyond meeting the requesters’ key search terms.
In the court’s October 12th Memorandum Opinion, U.S. District Judge Christopher Cooper articulated the main issue with the methodology of performing a FOIA search when he alluded to common FOIA request processes as being like the board game “Battleship.” In this scenario, the requester is expected to “score a direct hit on the records sought based on the precise phrasing of his request.” Cooper elaborated the process by claiming that “FOIA cases are typically resolved on summary judgment,” emphasizing that often the end goal is not specifically the accuracy of a search but the way in which a search is conducted.
In a November 8th article titled “Machine Learning Technology Can Help Agencies Handle New FOIA Request Standard,” Jon Kerry-Tyerman discusses a machine learning technique known as “clustering,” which can be beneficial in identifying a multitude of terms in files relevant to a FOIA request by understanding language patterns. According to Kerry-Tyerman, “clustering technology will organize documents into high-level themes,” which will expedite the process of discovering and distinguishing relevant items from irrelevant ones. While clustering is a relatively simple process, “predictive coding” is a more advanced procedure that agencies can use to train machine learning tools to perform these kinds of searches with supervision.
AINS understands the benefits that machine learning can provide both FOIA processors and FOIA requesters. Our FOIAXpress Electronic Document Review (EDR) Module uses machine learning to cluster documents based on key words, themes, and other attributes. Not only does it speed document review, but it also improves responsiveness, as more relevant documents are delivered to the requester. In addition to document clustering, EDR also automatically de-duplicates emails, consolidates email threads, and ranks and categorizes documents according to search terms. Our EDR customers have seen an up to 60% reduction in time spent on document review.