The purpose of this study is to develop a methodology as to the knowledge management of concept structure for learners. Fuzzy clustering is adopted to implement classification so that learners of the same cluster have homogeneity and display common features of cognition diagnosis. In addition, fuzzy clustering is based on information of concept scoring and caution index from polytomous student-problem chart. In the study, the empirical data is the assessment of statistics concepts from university students. The results show that there are four clusters and each cluster has its own cognitive characteristics. To sum up, the methodology can improve knowledge management in classroom more feasible. Finally, some recommendations and suggestions for future investigations are also discussed.