Clustering and Recurring Anomaly Identification: Recurring Anomaly Detection System (ReADS)

This viewgraph presentation reviews the Recurring Anomaly Detection System (ReADS). The Recurring Anomaly Detection System is a tool to analyze text reports, such as aviation reports and maintenance records: (1) Text clustering algorithms group large quantities of reports and documents; Reduces human error and fatigue (2) Identifies interconnected reports; Automates the discovery of possible recurring anomalies; (3) Provides a visualization of the clusters and recurring anomalies We have illustrated our techniques on data from Shuttle and ISS discrepancy reports, as well as ASRS data. ReADS has been integrated with a secure online search