A generic architecture for the development of intelligent programming tutors

Novice programmers experience a number of learning difficulties in grasping programming concepts. Research shows that individualised tutoring is the only means of assisting learners in overcoming these problems. However, both time and economic constraints make this infeasible. Intelligent tutoring systems have provided a solution to this problem. Studies conducted reveal that students using intelligent programming tutors were able to overcome their difficulties in a shorter time span compared to those students who did not use an intelligent programming tutor. In spite of this, the intelligent programming tutors developed thus far have certain limitations. Furthermore, the development costs, especially the time taken in designing and implementing such systems, are very high. There appears to be a need for a generic framework or architecture for the development of intelligent programming tutors. The main aim of this paper is to propose such an architecture. The paper firstly provides a brief account of the intelligent programming tutors developed thus far. A critical analysis of these intelligent programming tutors is then provided. Based on this analysis, a generic architecture for the development of intelligent programming tutors to teach the object-oriented and procedural programming paradigm is proposed. The paper describes the overall architecture, each module comprising the architecture, and the different artificial intelligence techniques used to represent information in each module. Finally, the paper examines future research that will be conducted in order to extend this architecture.