The Study And The Application On Algorithms Of Auto-Generating Examination Questions

Examination is the most important part of the teaching course. How to choose topics and the organization of examination papers are the key factors in deciding whether the examination is a success or a failure. With the rapid development of computer technology, network communication technology and artificial intelligence technology, and with the gradual popularization of computer in the field of education, the demand of Auto-Generating examination questions with computer technology becomes more and more urgent.From the point of view of algorithms and mathematical models, Auto-Generating examination questions are mathematical problem of multi-objective optimization in a certain Constraint condition. From the point of view of the approach and solution, Auto-Generating examination questions are problems of a library of questions design as well as pumping algorithm design. Using the traditional method to solve such problems, there are some shortcomings such as slow speed and low quality of Generation examination questions, the low success rate, etc. In view of above insufficiency, this text solved this problem by the genetic algorithm that solves the complex problems of random search and optimization by simulating biological genetic variation and natural selection mechanism, and compared the characteristics of several commonly used genetic algorithm. Combining the specific requirements of topics selection and organization in difficulty, distinction and Knowledge-point coverage indicators, this text improved in many aspects including the establishment of a database of examination questions, and determining the initial population, encoding scheme, fitness function design, selection, crossover and mutation, as well as design iterations and termination conditions, introduced the improved genetic algorithms. The design of the fitness function tends to be more concise and reasonable, being able to achieve the goal that the results of selected topics meet the needs of people more. For the design of selection, cross and mutation, we adopted dynamic design, eliminating the illegal population and the low value of the fitness function of chromosomes as soon as possible in order to effectively shorten the time for generation examination questions. And the Comparison with a random selection method shows that the improved genetic algorithms in this article have obvious advantages in the fitness function value and stability, being able to meet the needs of users to the largest extent. It is scientific, rational and better practicality.This text designed and realized the system of Auto-Generating Examination questions base on genetic algorithm, adopting JSP programming technology, dynamic web design technology and database technology, and carried out functional testing as well as performance testing.