Cognitive scaffolding for a web-based adaptive learning environment

On-line Web-based learning environments with automated feedback, such as WebLearn [5], present subject questions to the student and evaluate their answers to provide formative and summative assessment. With these tools, formative learning activities such as quizzes and tests are mostly pre-planned, since testing instruments are generated by selecting questions in a pre-specified manner out of question banks created for the purpose. Although this approach has been used with a significant degree of success, the real challenge to support students' learning is to mimic what a human instructor would do when teaching: provide guided learning. The main difficulty associated with creating such an 'electronic tutor' is to implement the required intelligent dynamic behaviour during learning. That is, at any stage of a student's learning session the system should take into account his/her demonstrated cognitive level to generate the next appropriate formative testing instrument. For students to be able to make the higher-level cognitive contributions as they progress through a session, the system must keep a history of students' answers and must react accordingly. We call here that behaviour adaptive learning by adaptive formative assessment. We propose on this paper a strategy to implement an adaptive automated learning system, based on establishing an incremental cognitive path from the lowest to the highest level questions related to a concept. In the research literature this has been often called 'cognitive scaffolding'. For our on-line automated environment, the first hurdle has been how to define the scaffolding and how to implement it from question banks that have not been created for this process. Our approach is embodied in WebTutor, a black box' component being developed at RMIT University to work in combination with the generation, presentation and feedback capabilities of the WebLearn system.