Abstract A system used to help Chinese language teaching activities at the primary school is proposed. This study aims to establish a bayesian network based adaptive testing (BNAT) and a digital individual remedial teaching language which is adaptable to BNAT. It helps teachers with individual remedial instruction. We hope it can reach significant effect on testing students’ individual abilities and meeting students’ individual needs. The focus of this study is on the unit of essay . After analyzing the content of the textbook, the structures of the content were constructed, and items were designed according to them as well. After the pre-test, ordering theory is used to decide the students' knowledge structure and those parameters were used in the item bank. Additionally, the remedial instruction is maked up by referring to the knowledge structure of experts and students. And the digital individual remedial teaching material is developed after that. The material was provided for experimental teaching. This experimental teaching includes three kinds of instruction models for experimental teaching. They are as follows: one teacher teaches one students, one teacher teaches two students, and one teacher teaches a group of students. Some findings are briefly outlined as follows: 1. The number of items administrated by BNAT is 13, 4 items are omitted and prediction accuracy can reach 98％. 2. After taking the adaptively remedial instruction, students have significant progress on their average grades. （one teacher teaches one students：t=2.07，p=0.047；one teacher teaches two students：t=2.04，p=0.049；one teacher teaches a group of students.：t=2.87，p=0.007） It shows that this digital individual remedial teaching material is valuable for using. 3. This digital individual remedial teaching materials making by knowledge structure is adaptable to three kinds of teaching models. 4. In point of the students for the low mark group, the teaching model of one v.s. one and one v.s. two are better than one v.s. a class. In point of the students for the high mark group and the average mark group, the three kinds of teaching models have no significance difference.
