A multi-agent approach to adaptive learning using a structured ontology classification system

Diagnostic assessment is an important part of human learning. Tutors in face-to-face classroom environment evaluate students’ prior knowledge before the start of a relatively new learning. In that perspective, this thesis investigates the development of an-agent based Pre-assessment System in the identification of knowledge gaps in students’ learning between a student’s desired concept and some prerequisites concepts. The aim is to test a student's prior skill before the start of the student’s higher and desired concept of learning. This thesis thus presents the use of Prometheus agent based software engineering methodology for the Pre-assessment System requirement specification and design. Knowledge representation using a description logic TBox and ABox for defining a domain of learning. As well as the formal modelling of classification rules using rule-based approach as a reasoning process for accurate categorisation of students’ skills and appropriate recommendation of learning materials. On implementation, an agent oriented programming language whose facts and rule structure are prolog-like was employed in the development of agents’ actions and behaviour. Evaluation results showed that students have skill gaps in their learning while they desire to study a higher-level concept at a given time.