The Research and development of Digital Teaching Materials and Evaluation Based on Knowledge Structure and Bayesian Networks-taking Cylinder Volume unit of The sixth grade as an example

[[abstract]]The purpose of this research is to establish a knowledge structure and Bayesian networks based mathematical teaching materials and computerized adaptive testing. Sternberg(1997) pointed out, the thinking styles of students are measurable. Thus, this study was to examine learning difficulties and analyzed the differences among various situations in math for students with differential thinking styles. First, we analyzed the curriculum based on the knowledge structure and edit the evaluation items according to the construction. Then, establish mathematical materials of experimental teaching. Applying the Bayesian Network to infer the mistaken type and skill as the reference of remedial instruction. In the same time, students were required to fill in thinking style scale form. After the experiment, analysis the data. Some finding are briefly outlined as follows： 1. After taking the experimental teaching, experimental group students; average grades were better than comparison group’s significantly. It shows this teaching material is valuable for using. 2. After experimental group students taking the remedial instruction, they have significant progress on their average grades(post-test > pre-test, 90.16>83.59).It shows this teaching material is valuable for remedy. 3. The rate of saving items by KSAT is above 58.96％ and prediction accuracy can reach 93.62%. 4. Analysis the regularity of error types and sub-skill under the CAT and full response. The regularity of pre-test is 97.96％ ,and the post-test is 98.70％ 5. The experimental group students with legislative styles have higher average grades in Cylinder Volume unit than those with judicial styles and executive styles. 6. Distinguishing the error type and sub-skill performances based on different thinking styles (1) Error made by the experimental group student with legislative styles was Less than those with judicial and executive styles in error type 14. (2) The experimental group student with legislative styles with sub-skill 14 and 17 was more significant than those with judicial and executive styles. (3) The experimental group student with legislative-globe styles with sub-skill 14 was more significant than those with judicial and executive style