Font Size: a A A

Study And Implement On Building Test Paper System Based On Genetic Algorithm

Posted on:2009-03-28Degree:MasterType:Thesis
Country:ChinaCandidate:L XiaoFull Text:PDF
GTID:2178360245980273Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
The examination is a mainly measurement of testing a student's knowledge and ability level. A examination should use educational surveying, psychometrics, educational statistics and computer science integrated and so on. It should have a scientific, objective and fair attitude on designing the test questions. The existing item bank managing systems mostly have no unified format standards, no effective sharing and no better method on auto-generating test papers. They cannot be used efficiently on the process of teaching. Thus it is necessary to integrate the existing item bank managing systems and study better algorithms on auto-generating test papers.At first, this article analyzed the exiting item banks' development status. Then, according to the examination's actual requirements designed the system's developing structure and scheme. It based on the Web and had the mainly functions of questions' management, generating test papers, graphic elements' management and submitting test paper. The Five Chapter detailed instroducted the auto-generating test paper. It analyzed the multiple generating test papers' conditions detailed and according to the actual requirements, utilized the genetic algorithms on the auto-generating test paper. Genetic algorithms had the features of multi-target parameters optimization and intelligent searching capability. Summarized the existing systems' characteristics and deficiencies, the auto-generating test paper model was based on the satisfying conditions of the test paper's structure, the total scores and the test questions' exposure rate. It used the difficulty proportion distribution and the coverage degree as the target requirements, and the test time requirement, test questions' reliabilities as the model's constraint conditions. Then, it used penalty function method on the regulation infeasible solutions and kept the diversity of the population. The Part Six introduced the environments of the system and analyzed the difficulties: introduction into the questions, the questions' repeatability detection method and the system's security. At last, the summary and prospect for future work were made.The system was proved having the universality, expansibility and security throught the practical operation. It could satisfy the complex generating test paper's demand. It was helpful to the implementation of the controlling quality, improvement of the teaching efficiency and carrying separation between teaching and testing.
Keywords/Search Tags:examination, auto-generating test paper, genetic algorithm, test paper model, penalty function
PDF Full Text Request
Related items