Auto Diagnosing: An Intelligent Assessment System Based on Bayesian Networks IEEE 2007 Frontiers In Education Conference- Global Engineering : Knowledge Without Borders, Opportunities Without Passports Liang Zhang, Yue-ting Zhuang, Zhen-ming Yuan, Guo-hua Zhan
Outline Introduction Architecture of the system Authoring module Training/test module Monitor module Grading module Knowledge map Diagnosing learning status Result and discuss Conclusion and future work
Introduction E-learning system has become more and more popular. Many effective assessment systems have been proposed. Conventional test systems simply provide students a score, and do not provide adaptive learning guidance for students.
Intelligent Tutoring Systems (ITS) Adaptive learning. Difficult and time consuming to assess student ’ s knowledge level or learning status for the teachers manually.
Architecture of the system Authoring module teachers can use to write their assignments or questions Training/test module designed mainly for student ’ s client. Monitor module used by instructors to keep track of student ’ s status. Grading module assesses student ’ s knowledge map.
Authoring module Manage question storage and make the schedule of a test. Question storage is composed of the questions, answers, evaluation criteria, degree of complexity, and difficulty. Relation strengths between concepts and the questions.
Training/test module Web-Based online training/test module is designed mainly for students. Features Client side control Time control Security control Auto-installation
Monitor module The real-time monitor module keeps track of student ’ s registration, submission and performance. Feedback including score, missing concepts, and next step help.
Grading module Use the fuzzy match algorithm. Automatically grade student ’ s answers, discriminate understanding or misunderstanding concepts of students. Finally, we use rule inference method to create learning guidance for the learner.
Knowledge map
If W1=0.3,W2=0.1,W3=0.6, the conditional probabilities of sub-section1
Diagnosing learning status Nodes represent student ’ s answer (right or wrong). BNs can absorb the evidence when students answer a question.
Learning guidance Stage1 :calculates degrees of P. Stage2 :select the max subjection degree and sends it to students.
Giving advice of next step EX: C1 and C2 are the prerequisite concepts of C3. If G is less than predefined threshold value.
Result and discuss
Conclusion and future work In this paper, presented an integrated approach to diagnose student ’ s learning status and provide learning guidance. In the future, technology of student modeling is worth studying deeply to improve the accuracy of knowledge map representation.