KBS-DetectOR: AUTOMATING ANOMALY DETECTION IN HYBRID KBSS

Authors:

• P. R. Stiger, University of Tulsa, stiger@euler.mcs.utulsa.edu
• R. F. Gamble, University of Tulsa, gamble@euler.mcs.utulsa.edu
• H. L. Christensen, University of Tulsa
• R. T. Plant, University of Miami, rplant@umiami.ir.miami.edu

Abstract

This paper presents details on the construction of KBS-DetectOR, an automated anomaly detection system for hybrid knowledge based systems (KBSs) that rely on objects and rules. A comprehensive, language-independent tool is needed as complex KBSs are developed and possibly embedded into other software systems. This paper concentrates on the interleaving of two detection mechanisms for hybrid KBSs into KBS-DetectOR. The tool encoding of previously defined algorithms for hybrid KBS anomaly detection is discussed and the interface of the tool is shown. The addition of other detection mechanisms is also discussed.