Research And Implement Of Automatic Evaluation Algorithm For Course Of Visual Basic

As a beginners'computer language course commonly used in colleges today, Visual Basic cries out for an automatic evaluation tool for programming to effectively improve the teaching quality and the teaching setup of the course. In order to achieve automatic evaluation in a course like Visual Basic, one of the keys lies in finding a scientific evaluation algorithm.Based on the study of available evaluation methods, the author of this article proposes a new evaluation algorithm, in which evaluation is broken down into two parts:interface evaluation and code evaluation. In interface evaluation, the control information is obtained by directly reading the window file of the evaluated program, and by comparing the information with evaluation standards, a grade is generated. The code evaluation is a combination of dynamic evaluation and static evaluation. In dynamic evaluation, the evaluated program is run on certain logics by simulating Windows messages, meanwhile the running features of the program are obtained by embedding temporary codes, a comparison of which features gives a grade. In static evaluation, the program is divided into several grading points in line with the evaluation standards, each grading points restricted by a regular expression, and then a grade is obtained by matching the regular expressions in the program codes.In order to assess the effect of the above evaluation algorithm, the author's team designed a prototype of the evaluation system, built a programming item pool and selected examples from students' program work. The process of the automatic evaluation experiment was recorded. Compared with manual evaluation, the errors of automatic evaluation lie within controllable range, its effectiveness widely acknowledged by teachers and students.The automatic evaluation algorithm studied in this article is independent of the programs, which may contribute to the enrichment of item pool. Nevertheless, this algorithm relies much on the evaluation standards, and the standard for each item can only be made manually by the creator assisted by the system. The automatic generation of evaluation standards is an object worthy of study in the future.