Experts and learners organize knowledge into networks of knowledge bits (nodes) which are interconnected by relational links. This paper discusses a network model used by teachers and learners for a knowledge domain (say thermodynamics) consisting of knowledge nodes and links like curriculum and course structures and links. Course structures tend to first focus on knowledge clumps (say ideal gases) and individual nodes (say gas constants) and the links interconnect these nodes in a forward-directed, prerequisite manner. On the other hand, novice learners focus on individual nodes (say definition of entropy) and then look back for links to prerequisite nodes (say definition of heat transfer) as they build their own knowledge networks. Expert forward-directed network search strategies are compared to novice backward-directed strategies in this paper. The goal of an expert AI search is to find a problem solution given inputs to the network while the goal of the learner is to find the perquisite knowledge required to master a knowledge node. Expert search strategies are then amenable to expert system AI strategies such as Bayesian statistics, cosine similarity and fuzzy logic. Learners seek to master the individual nodes and the relational links between nodes so that they can construct their own expert knowledge network. The strategies for meeting the learner goals are then completely different from those of the expert and not suited to well-known AI searching methods. This paper also discusses the requirements of an AI system designed to assist the learner in meeting their goals as efficiently as possible. A brute force AI system which assists learners with building links and their knowledge network is discussed in this paper.
|Journal||ASEE Annual Conference and Exposition, Conference Proceedings|
|State||Published - Jun 22 2020|
|Event||2020 ASEE Virtual Annual Conference, ASEE 2020 - Virtual, Online|
Duration: Jun 22 2020 → Jun 26 2020