Widening the Knowledge Acquisition Bottleneck for Constraint-based Tutors
Intelligent Tutoring Systems (ITS) are effective tools for education. However, developing them is a labour-intensive and time-consuming process. A major share of the effort is devoted to acquiring the domain knowledge that underlies the system’s intelligence. The goal of this research is to reduce this knowledge acquisition bottleneck and better enable domain experts with no programming and knowledge engineering expertise to build the domain models required for ITS. In pursuit of this goal we developed an authoring system capable of producing a domain model with the assistance of a domain expert. Unlike previous authoring systems, the Constraint Authoring System (CAS) has the ability to acquire knowledge for both procedural and non-procedural tasks. CAS was developed to generate the knowledge required for constraint-based tutoring systems, reducing both effort and the amount of knowledge engineering and programming expertise required: the domain expert only has to model a domain ontology and provide example problems (with solutions). We developed novel machine learning algorithms to reason about this information and thus generate a domain model. A series of evaluation studies have produced promising results. The initial evaluation revealed that the task of composing the domain ontology aids the manual composition of a domain model. The second study showed that CAS is effective in generating constraints for non procedural database modelling and the procedural data normalisation. The final study demonstrated that CAS is also effective in generating constraints when assisted by only novice ITS authors; under these conditions it still produced constraint sets that were over 90% complete.