Short answer question examination using an automatic off-line handwritting recognition system and the Modified Direction Feature

Abstract

Handwriting Recognition is one of the most intensive areas of study in the field of pattern recognition. The automatic assessment of exam scripts [1] can benefit from off-line handwriting analysis methods. Off-line automatic assessment systems can be an aid for teachers in the marking process; it reduces the time consumed by the human marker. Recently developed systems in this area have achieved an encouraging assessment yield range executed under constrained conditions. There has not been any recent work in the development of off-line automatic assessment systems using handwriting recognition, even though such systems will clearly benefit the education sector. There is a significant relationship between feature extraction and the recognition response yield; therefore the current research proposes the use of the Modified Direction Feature (MDF) extraction technique [2] and neural networks to develop an automatic assessment system for marking short answer questions. The system has high assessment accuracy (up to 92.31% for hand printed and 91.67% for cursive handwritten). The proposed system also includes marking criteria to augment its accuracy.