Аннотация:In a testing system, the quality of the test items will significantly affect the accuracy of a test; therefore, several measures have been proposed to represent the quality of each test item, e.g., degree of difficulty and discrimination. However, the quality of a test not only depends on the quality of the item bank, but also relates to the way the test sheet is constructed. Selection of appropriate test items is important when constructing a test sheet that meets multiple assessment requirements, such as average difficulty degree, average discrimination degree, length of the test time, number of the test items, and the specified distribution of concept weights. In this paper, a multiple criteria test-sheet-generating problem is formulated and a dynamic programming approach is proposed to cope with the problem. The new approach utilizes the techniques of clustering and dynamic programming, enabling construction of a feasible test sheet in accordance with specified requirements. Some experimental results of 20 cases show that the new approach is capable of achieving better performance than previously used test-sheet-generating methods.
