Research On The Design And Practice Of Genetic Algorithm Test Paper Composition For Junior High School Students's Mathematical Literacy Based On PISA

As a literacy assessment project that emerged from OECD and targeted at 15-year-old students,PISA has attracted the attention of more and more experts and scholars in the education field,and the results of related research in the field of mathematics are also continuing increase.With the development of information processing and measurement technology,there has been a precedent for the computerization of the PISA,that is,PISA is carried out in the form of computer-based testing.For the time being,when it comes to the research of computer testing,it will involve a series of questions such as question bank and test paper composition.Due to the international background of the PISA,it is a big challenge for researchers to achieve the localization of the PISA on the basis of adapting to domestic conditions.Therefore,the research adopts the literature research method,design-based research method and mathematical statistical analysis method,selects the mathematics field in the PISA,combines the content of the Fujian Provincial Mathematics High School Entrance Examination,focuses on the design of a paper generation algorithm based on the genetic algorithm idea,and verifies the quality of the algorithm through practice.In this research,the main content of the author's work is as follows:First,the thesis analyzes the current research status of PISA mathematics literacy and automatic test paper generation,determines the issue to be solved in the research,tries to identify the significance of the research,and preliminarily plans the research methods to be used.Subsequently,the thesis starts from the research issue,analyzes the definition of literacy from etymology,distinguishes literacy and quality,expounds the different definitions of mathematical literacy,and sorts out the theoretical basis involved in the research.Secondly,the research combs the formation model and evaluation framework of mathematical literacy,and combs nine indicators of mathematical literacy measurement.After that,five basic principles that ought to be observed in generating test papers are proposed,the content of the test paper assessment is determined,the essential cognitive requirements are sorted out,the main attribute parameters of the test questions are analyzed,the constraints of the test paper are determined,and the corresponding mathematical model is built.Furthermore,after comparing and analyzing the merits and drawbacks of the five common algorithms for generating test papers,the idea of genetic algorithm is selected for research,and the key content of the algorithm is designed,the following determined respectively:(1)chromosome coding method: use the question type as segments,segment real number coding;(2)selection operator: roulette selection;(3)crossover operator: segment crossover,including single point crossing and whole section crossing;(4)mutation operator:segment single point mutation.Meanwhile,the research constructs the objective function and fitness function suitable for test paper composition,and considers incorporating the idea of elite retention in the algorithm.Finally,the research applies the designed algorithm to generate ten test papers,selects the one with the highest adaptability,communicates with some front-line teaching backbones of mathematics subjects in a middle school in Zhangzhou,revises some contents of the test paper,and then select class X of grade 9 to test,and conducts a matching questionnaire survey on the remaining eight dimensions of mathematics literacy,uses SPSS22.0 to perform statistical analysis on the test score data.The results show that the test scores are in a quasi-normal distribution,the actual pass rate of each question on the test paper is roughly close to the preset difficulty,which proves that the quality of the test paper is good and the algorithm designed by the research can basically meet the requirements.This research believes that the main purpose of designing the automatic test paper generation algorithm for mathematical literacy assessment is to improve the convenience and effectiveness of mathematical literacy assessment.So far,the algorithm designed by this research can basically meet the test requirements and has some practical value.In order to better adapt to the new situation of personalized learning and adaptive assessment,the follow-up research may try to further improve the algorithm from some aspects such as the test paper generation system and test parameters.