The Research On The Mathematic Model And Algorithm Of Intelligent-generating Test Paper

Compared with traditional examination form, the Computer Aided Test system canimprove the efficiency and quality of examination own to the information technology. As keypart of the system, intelligent-generating test paper strategy has an important impact on theperformance of the entire system. Traditional strategy does not establish an appropriatemathematical model. The running time and quality of the algorithms such as randomalgorithm and backtracking algorithm are not as good as people planned. In recent years, theperformance and quality of the intelligent-generating test paper is particularly importantbecause of the exponential growth of test paper resources.In this thesis, the main task of intelligent-generating test paper strategy was analyzedbased on the theory of educational measurement. Then the common mathematical models andalgorithms were introduced. After analyzing the advantages and disadvantages of the differentmodels and algorithms, the body of the intelligent-generating test paper includingmathematical model and the improved genetic algorithm was established. The mathematicalmodel is a multi-objective optimization function contains different weights quality subfunction and common model constraints. The mathematical model covers different forms andtypes of the examination.To avoid the long running time, uncontrolled quality and testing defects of someintelligent-generating test paper strategy, the encoding scheme, genetic modification andgenetic parameter of the genetic algorithm were improved and used in the mathematicalmodel. The simulation analysis on experimental data of the model and algorithm indicatesthat the running time is fast and the quality is controlled by plan. No failed cases are found inthe whole running results of the simulation.