| Students' homework and its correcting are very important teaching process in college education. A great deal of research indicates that the behaviors of both students and teachers have effects on students' studying qualities. Compared with homework corrected manually, the application of computer technology to students' homework correction can greatly improve homework correcting speed and reduce teachers' workload. Now there have been many teaching programs for homework correcting, but most of them can only correct the objective questions, such as multiple-choices and true/false questions, while many subjective questions such as computation questions with multiple-steps, which are widely used in the basic courses for science and engineering specialities, are not acceptable. A method of automatic recognition and correcting of some subjective problems by computers is presented in this paper. Using symbolic computation and expert system technology, the recognition technology of elementary mathematical expressions and applies this method to the course of Material Mechanics is studied in this paper.The automatic correcting of subjective questions is a knowledge based recognition technology. The knowledge representation which is especially suitable for question correcting is proposed in this paper. The knowledge in this domain is divided into theoretical course knowledge and expertise course knowledge. The effective structure of knowledge representation must be designed to organize and represent theoretical course knowledge in order that computers can understand and correct student's answer. A solution step in theoretical course knowledge is nest expressed using the arithmetic operators such as SUM, TERM and FIELD, some complicated FIELD arithmetical operators simplified by variable O which can improve the accurateness and effectiveness of homework correction. In order to describe the mistakes and problems examined from the student's homework, the course expertise of experienced teachers is involved in the process of correction. The mistakes and problems found in the students' homework, which are useful for student's further study after homework, are described by defining key steps and error steps.The comparative recognition of mathematical expressions is the key part ofhomework correction. Using symbolic computation technology such as computation of maximum common divisor and factorization, the comparative recognition of some elementary mathematical expressions is realized in order to lay a technical foundation for subjective problem's automatic correction.A reasoning control strategy is also proposed. The process of correcting to subjective problems is relatively complicated. The whole process is divided into four stages: pre-processing, recognition of students' right answers, recognition of students' wrong answers and recognition of students' similar wrong answers. There are three correcting steps adopted in the whole process except the pre-processing stage: multi-solution recognition, single-solution recognition and solution step recognition in order to make the whole distinguish more precise and reliable.The automatic tracing and dynamic diagnosis of errors in students' homework is also implemented in the process of homework correction. After student finishing his homework on a computer, the computer can diagnose, records and reports the mistakes and problems appearing in his homework, according to which the student can study more pertinently and the teacher can adjust his teaching plan.This research was sponsored by World Bank loan project of the China Education Ministry. The application of the homework correction system to the course of Material Mechanics used for two terms in our university indicates that the automatic correction and correcting technology is relatively steady and reliable, and the reform scheme against traditional homework measure is feasible, and the effectiveness of improving correlative course's holistic teaching quality is remarkable.
|
PDF Full Text Request