Recognising irregular answers in automatic assessment of fill-in-the-blank tests

Student assessment in Learning Management Systems (LMS) is mostly conducted by means of close-ended questions that are simple to evaluate without human intervention. Fill-in-the-blank tests are usually more challenging since they require the test-taker to recall concepts and associations not available in the question' statements. Automatic evaluation of the latter will give correct credit to the test-taker only when the response is entered verbatim, that is, free of orthographical or typographical errors. In this paper we consider a straightforward application of a text-matching algorithm that may prevent missing scores in automatic fill-in-the-blank assessment, when irregular (similar but not exact) answers are entered because of those types of errors. A small experiment with an email security questionnaire is presented showing the potential of the technique.