John C. Ownby
SAX is a diagnostic system designed to search for inventory related errors in a large transaction base, propose solutions for correcting the errors, and extrapolate the identified errors into patterns of behavior. SAX uses a modular combination of backward-chaining rules and mathematical algorithms to replicate domain knowledge. The system was developed in two phases, with the error seeking phase deployed in 1990 and the pattern recognition phase in 1992. Today, the complete system is in production and provides expert diagnostics to 13,000 salespersons, 1,000 sales managers, and 60 clerical accounting employees.