EFFICIENT COMPUTERIZED TESTLET-BASED TEST GENERATION UTILIZING DPSO WITH PARTIAL COURSE ONTOLOGY

摘要

This paper argues that how to formalize test requirementsrndynamically and how to select suitable items effectivelyrnare two main challenges in the current researches ofrncomputerized test generation. The proposed approachrnobeys the Discrete Particle Swarm Optimization (DPSO)rnscheme and uses partial course ontology to generate testsrnaccording to the studied materials. Besides, it alsornharnesses the notion of testlets to construct testlet-basedrntests for students. In the experiments, the results show thatrnthe proposed approach can effectively formalize the testedrnconcepts and their weights, and also generate the wellscrutinizedrntest in accordance with the test requirementsrnin the acceptable time.