Text Mining of Maintenance Data

Every year the US DoD collects millions of maintenance and repair records on everything from leaky faucets to helicopter transmissions. The records often contain free-text descriptions of problems and how they were resolved, complete with the expected typos, abbreviations, and misspellings typed by busy technicians.

SmartArrays developed a tool to assist analysts in categorizing US DoD maintenance records related to corrosion, as part of a Congressional mandate to reduce corrosion effects. The tool allows analysts to explore the characteristics of free-form text records, develop rules to identify costs resulting from corrosion, and process data records to identify corrosion-related maintenance. The text analysis facility of this tool allows it to be used in a variety of other contexts besides the initial corrosion analysis.

Analysts originally ran SQL stored procedures that required hours to execute in order to analyze the text records. The SmartArrays-based tool runs on a desktop and provides results a few seconds.