Polytomous knowledge structure and assessment automaton under competence model

The knowledge space theory has provided new methods and approaches for conducting knowledge tests and assessments for examinees. At present, knowledge assessment based on knowledge space theory has been widely applied to deterministic finite automata. However, most of the current automata used for knowledge assessment consider binary response values for items, and the assessment of knowledge states is limited to conjunction models and disjunction models, without considering more generalised competence models. This obviously restricts the application of automata in the field of computerised adaptive testing and assessment. This paper combines the competence-based knowledge space theory with the attribute multi-mapping method and proposes two questioning rules, the medium difficulty questioning rule and the difficulty-adaptive questioning rule, according to the difficulty level of the items. Furthermore, two assessment methods of polytomous knowledge states and competence states under competence model are proposed. Finally, this paper validates the feasibility of the polytomous knowledge structure assessment automaton under competence model through empirical research and verifies through simulation experiments that the two questioning rules of the polytomous knowledge structure assessment automaton have a beneficial effect on the performance improvement of the automaton, as well as the main factors affecting the difficulty-adaptive questioning rule.