C-language syntax tutoring using machine learning techniques

Abstract

This study describes a prototype system that creates a rulebase by capturing compiler error messages and uses them as a means for building a decision tree. The rulebase is then used to assist novice programmers with cryptic C-language compiler error messages. The system, called the C-Compiler Error Message Interpreter (C-CEMI), generates a data file which is used by an inductive learning system known as C4.5 to discover the rules. The rulebase is incorporated into a dynamic C-language grading system called Ceilidh.