Strategies for Tutoring Multiple Bugs
Abstract
This report investigated a problem in the field of intelligent Tutoring Systems that has not been effectively addressed before. That problem is how to help students who need help with more than one error, or bug. For example, novice students who write computer programs rarely make a single bug. Rather, they come to their tutors with five, six, seven, or more bugs and they need help with all the bugs. Empirical observations of experienced human tutors show that they do not just jump in and start tutoring the first bug in the program. Instead, tutors formulate a tutorial plan for helping the student. Creating the tutorial plan requires the tutor to answer the following five tutorial planning decisions about each bug. This research studied the problem of multiple bugs. Starting from empirical observations of experienced human tutors, a model was developed of the knowledge required to answer the five tutorial planning questions. The model is implemented as a computer program, TP, that can develop tutorial plans for multiple bugs that are as good as the tutorial plans generated by experienced human tutors. Keywords: Cognitive psychology; Intelligent tutoring; Automated diagnostic inference; Computer programming; and Instruction.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1990
- Accession Number
- ADA223921
Entities
People
- David C. Littman
Organizations
- Yale University