Confident assessment of children's handwritten responses
Tóm tắt
This paper introduces a novel approach for the automatic assessment of children's responses to standardised English exam questions. The constrained nature of the question and answer medium is exploited to produce an automatic assessment mechanism that is both highly accurate and produces a reasonable level of response yield. It is shown that the novel approach can achieve 100% scoring accuracy on 44% of all responses compared to a traditional lexical approach that has an error rate of 41%. When a thresholding method, similar to that used in the novel approach is applied, the traditional approach can achieve an accuracy of 100% but with a response yield of only 5%. The approach introduced in this paper is thus shown to have a significant advantage over the traditional lexical based assessment.
Từ khóa
#Iris #Error analysis #Automatic testing #Aging #Publishing #Conferences #Handwriting recognition #Maintenance #Shape #WritingTài liệu tham khảo
payne, 1999, Human motor development; A life span approach, 4th ed, Chap 12
luttinger, 2001, Learning disorder: Written expression, eMedicine Journal, 2
10.1109/ICDAR.2001.953921
dehkordi, 2000, Case classification of off-line hand-written words prior to recognition, IAPRS
vermeuelen, 1995, A comparison of hmm and neural network approaches to real world telephone speech applications, IEEE International Conference on Neural Networks and Signal Processing, 796
10.1109/34.730549
evans, 1999, Holistic recognition of static handwriting using structural features, Document Image Processing and Multimedia (DIPM'99) IEE Colloquium 99/041, 121