Zachary A. Kudlak and LT Justin P. Sherman, U.S. Coast Guard Academy

Abstract: When someone is lost at sea, the US Coast Guard coordinates a search and rescue (SAR) response. After-action reports (AARs) from these incidents may contain vital information about a person’s survival ability. Cadets and faculty from the US Coast Guard Academy used text analysis tools, natural language processing (NLP), and other methods from machine learning to identify and analyze SAR AARs about people in the water (PIW). This experience demonstrates how faculty can expand the horizon of their research program, include students in knowledge discovery, and produce tangible results. A classification model was developed to label SAR AAR files according to the presence of a PIW. The authors and their students continue to enhance the model and apply it to other Coast Guard-related domains with text-based reports.