Text Mining for Student Assessment

A portion of the University General Requirements Unit’s program is dedicated to instruction in English writing. Student assessment in the writing program generates a large amount of marking for instructors. A number of research and commercial packages have been identified that could assist in automated grading, but until recently the expense and effort required to use them seemed too great.

A number of methodologies have been used by previous researchers in automated grading of essays. The literature does not appear to include any references to the use of Classification Association Rules Mining (CARM) to the problem of automated essay grading.

This project extends UGRU’s existing proprietary suite of computer-based assessment tools to collecting and grading of essays. It then investigates the applicability of CARM techniques to assigning a coarse grade to the essays using commercially available data mining tools. Mining models based on strictly the text submitted by students, on strictly linguistic attributes of the text, and on a combination of the two sets of attributes were examined. Evaluation of the mining classification shows that it is possible that CARM techniques could be usefully applied to the problem of automated grading of student essays.