Abstract: | Detecting that a student has made an error is typically easier than accounting for the error. Determining any underlying beliefs that contribute to the occurrence of an error is a difficult business, both for the student and for the teacher. In automating this process, we face some further difficulties. We are in the process of building a system, CARoM, that is to be capable of diagnosing student's misconceptions within the domain of building simple electrical circuits. Eventually, we hope to exploit this information in a larger system. The motivation for this paper stems from a need to assess what has been achieved and the limits of what can reasonably be achieved within an essentially constructive approach. We therefore focus on those issues in automated diagnosis that have a direct bearing in the context in which we are trying to provide improved diagnostic information. We review the methods that are relevant to diagnosing student misconceptions and briefly assess the prospects for further work. This requires reviewing research on the representation of misconceptions as well as that on diagnosis itself.
|