Usage of Graph Patterns for Knowledge Assessment Based on Concept MapsThe paper discusses application of concepts maps (CMs) for knowledge assessment. CMs are graphs which nodes represent concepts and arcs represent relationships between them. CMs reveal learners' knowledge structure and allow assessing their knowledge level. Step-by-step construction and use of CMs is easy. However, mere comparison of expert constructed and learners' completed CMs forces students to construct their knowledge exactly in the same way as experts. At the same time it is known that individuals construct their knowledge structures in different ways. The developed adaptive knowledge assessment system which is implemented as multiagent system includes the knowledge evaluation agent which carries out the comparison of CMs. The paper presents a novel approach to comparison of CMs using graph patterns. Graph patterns are subgraphs, i.e., paths with limited length. Graph patterns are given for both fill-in-the-map tasks where CM structure is predefined and construct-the-map tasks. The corresponding production rules of graph patterns allow to expand the expert's constructed CM and in this way to promote more flexible and adaptive knowledge assessment.

In concept map-based assessment an expert’s concept map can be expanded using graph patterns to add hidden and inverse relations. This helps to avoid forcing a learner to use certain structures and names. Graph patterns are subgraphs that describe combinations of concept map elements, from which extra relations can be inferred. In this paper an enriched set of graph patterns is described along with their respective IF...THEN rules which can be used for automated knowledge assessment. Some of them are already implemented in the intelligent and adaptive knowledge assessment system IKAS.

The paper is devoted to the concept map-based intelligent assessment system that promotes students' knowledge self-assessment and supports teachers in improvement of study courses through systematic assessment and analysis of students' knowledge on the basis of concept maps. During the last five years, both the system's functionality and knowledge assessment approach were improved persistently, and at the moment, certain level of maturity is reached in both directions. The paper focuses on general principles of functioning of the last prototype, tasks provided, teachers' and students' support, scoring and adaptation mechanisms. At the end of the paper, retrospection of the system's evolution and evaluation results is provided.

No abstract

Concept maps are an educational tool that is used both for learning and knowledge assessment. Automated knowledge assessment has a limited ability to assess non-standard constructions. and current automated concept map-based knowledge assessment systems fail to recognise the fact that learner's chosen labels for conceptual relationships may differ from teacher's preferred labels and still be correct. This paper conceptually introduces several mechanisms for overcoming this drawback which distinguish real misconceptions from correct statements that are expressed using different words.