|
With the development of the network tutoring environment, the
importance of the test and evaluation grows gradually. A
measured test is helpful for teachers to see if students digest
the knowledge and skills well. It is also helpful for students
to find out the bottlenecks of learning and hence achieve more
significant progress. The key to a good test is not only the
large database of test items but also the assessment strategy
which will have influence on the effect of the test and
evaluation. Because of the reasons above, we design an algorithm
which combine the concept of fuzzy theory, dynamic programming
and fuzzy ART to optimize the allocation of test items. Through
the algorithm, the allocation of test items will meet the needs
of considering multiple parameters, include of the degree of
difficulty and discrimination and the balance of the concept, to
reach the optimal allocation. Based on the algorithm, a network-
based testing and evaluation system has been implemented. Some
experiments have been done on the Nature science course of
primary schools. The experimental results show that our method
achieve satisfiable performance in allocating test items.
|