Assessment of Leaner's Understanding: An Experimental Result

Abstract

Assessing leaner’s answers is very time consuming for educators and limits them to be involved in other activities. In many cases, exam papers comprise of questions that require learners to write at least one or two sentences toe express their understanding. However, there are not many computer-based assessment tools due to limitations in computerized marking technology. Our research attempts to address this limitation by introducing a technique to evaluate short free text answer. It is based on a hybrid approach that combines natural language processing, information extraction and artificial intelligence. A textual answer is converted into a node link representation to extract the hidden knowledge structure. We then apply excess entropy to compute the amount of known information for each model and later compute the score accordingly.