Based On The Modified Random Algorithm Smart Group Volume System Design And Implementation

The essence of the computer intelligent generating test paper is solving the optimization problems of multi-objective parameter which will impact the quality of examination paper. Such as trace and trial algorithm, randomly selected algorithm, priority algorithm, genetic algorithm, error compensation algorithm, these are the algorithm used for automatically composing test paper. However, each algorithm has its advantages and disadvantages, so far there is no a good program to solute the algorithm of the automatic test paper. Therefore, designing a high-quality, accurate, fast intelligent test system is the research objective of this paper.In this paper, based on the analysis of a lot of domestic and foreign literature, firstly conducted a detailed analysis of the indicators of the optimization problem of the intelligent generating test paper, and pointed out the inadequacies of traditional mathematical models. Based on this topic, improved objective function and the mathematical model of the optimization problem of the intelligent generating test paper, taking the distribution of important difficulties, the level of knowledge and ability, the coverage of chapter, the types of questions, the quantity of questions, the degree of difficulty as the major controlling parameters, have been proposed firstly. Against to the traditional algorithms having the disadvantages of low success rate, slow convergence and quality, and so an improved algorithm, based on randomly selected algorithm, has been proposed to solve the optimization problem of the intelligent generating test paper.In addition, the subject found, the test paper auto-generation process is from the test question formation, to question database organization, until composition examination paper and analysis feedback’s periodic complex process. Analyzes based on above, in the process of research and development of the intelligence testing-paper generating system, we use the classical test theory as the theoretical basis of the construction of questions, at the same time, this paper study and analyze the inadequate of classical test theory in the determination of item difficulty coefficient, and learned from previous research experience, and proposed combining with the potential tendency theories to achieve item difficulty coefficient adaptive. This is another research focus of this study.This study eventually developed an intelligence testing-paper generating system by integrating the C#programming language and the technology of SQL Server database. Finally, using this system, implemented a number of testing-paper test. The experimental results show that this system can be in accordance with the test paper parameters, like the distribution of important difficulties, the level of knowledge and ability, the coverage of chapter, the types of questions, the quantity of questions,the degree of difficulty and so on, and quickly search the questions meeting the requirements, and seek the best test paper program, automatically generate high-quality papers, and successfully solve the problem of time-consuming generating test paper, low-quality examination paper.