Calculus is one of the most important courses in university mathematics education. Specially for students in scientific area, in order to prepare for future advanced course one need to lay a good foundation of mathematics skills. However in most Calculus curriculums’for teaching and assessment, frequently allow only paperandpencil examinations. Even thought a few examinations are computerized but still limited in the format of choice test. In stead of most “Computer NonAdaptive Testing” only providing the score but back coupling, this program is constructing a “Bayesian Network Calculus Computerized Adaptive Diagnostic Test” and “Learning System”. This system bases on the structure of competence indicators of selected mathematics topics and Bayesian network's deduction. It is adaptive and can discover the learning scotoma of student. We will also establish the Calculus construction response topics system. The significance of the system is to provide a tool where students may precede the problem solving on the computer system directly. Therefore, besides the data provided as traditional written examination, the system may also record how the students use the tools in system when solving problem. In other words, the information of answer response, tool, course, mathematical formula and time of student problem solving can be recorded to analyze student's erroneous picture. These will deduce the student ability and misconception. Finally, we will precede an experience test in class to verify the affectedness of the instruction.