| As an important part of the teaching process in our country, nowadays, the exam isstill a significant means of evaluating education. Computer Aided Testing (CAT) is bornby the development of the use of the means of the development of the information societyand educational measurement. Test-sheet composition is a multi-objective optimizationunder certain constraints, and is difficult to solve using the traditional mathematicalmethods. The efficiency and quality of the generated test-sheet depend on the testdatabase design as well as the algorithm used. More and more experts and scholars payattenion on test-sheet composition to find optimal test-sheet combination. And theresearch and development of test-sheet composition has yet gotten more and moreattention of a growing number of experts and scholars.A novel robust hybrid meta-heuristic optimization algorithm in terms of analytichierarchy process (AHP) and improved accelerated particle swarm optimization (APSO)is proposed for the characteristics and requirements in test-sheet composition problem.Firstly, according to the requirements of the test-sheet composition, a multi-constrainedmulti-objective model of test-sheet composition is created. Secondly, AHP is used towork out the weights of all the test objectives, and then the multi-objective model isconverted to the single objective model by the linear weighted sum. Finally, an improvedAPSO method is used to fulfill test-sheet composition. To prove the performance of theproposed method, it is compared with other population-based optimization methods. Theexperimental results show that the proposed method can greatly improve compositionspeed and success rate. Meanwhile, the pros and cons of the proposed method as well asdevelopment trend of the test-sheet composition are also discussed in this dissertation. |
PDF Full Text Request