ATTRIBUTE PATTERN-MATCHING EQUATIONS - A DIAGNOSTIC-TECHNIQUE FOR INTELLIGENT TUTORING SYSTEMS

The diagnostic process of an intelligent tutoring system is the proced ure by which the system determines a student's level of performance ba sed upon his or her problem solving behavior. In order for a system to insure the satisfactory performance of this process, three key issues must be addressed: (1) how to assign credit or blame to individual sk ills, (2) how to handle the problem of noisy data, and (3) how to hand le the problem of combinatorial explosion inherent in many Al applicat ions. In this paper we describe an informal reasoning technique, known as attribute pattern matching equations (APMEs), which is used in the diagnostic process of the TAPS tutoring system. This technique addres ses all three of the foregoing issues in assessing the performance lev els of individual skills in the TAPS system. In addition, it will be d emonstrated that APMEs are an effective method for reasoning with unce rtainty on three different levels.