The Research And Implementation Of Learner’s Classification For Chinese Teaching System Based On Fuzzy Clustering

Along with the world’s rising interests in Chinese, as one of the most promisingextension of traditional teaching methods, online teaching system has caught moreand more attention. Learner’s classification is a kind of processes, which divides thelearners of an online system into a number of different classes based on registeredproperties. There are more similarities between any two learners in a same class thanin different classes. Clustering is one of the most important branches in data mining.Traditional clustering is a kind of mechanical division, which strictly divides each ofthe individual objects to be identified into a class with the nature of critical definition.But in fact, the majority of learners do not have strict properties. Clustering combinedwith fuzzy mathematics, provides a new technology called fuzzy clustering. It is akind of soft breakdown, which is better corresponding to the real word.In this paper, we use an algorithm named “Clustering based on FuzzyEquivalence Relation” to implement the learner’s classification. At first, we calculatethe fuzzy similarity relation, and then the fuzzy equivalence relation, which isexpressed by transitive closure, the last; we divide learners into different classes bydifferent confidence level of the transitive closure matrix. Without using the ideaabout initial clustering center, this algorithm is prevented from falling into localminimum like FCM, and has the ability of global optimization. In addition, matrixoperation in this algorithm is suit for compeering.The implementation of fuzzy clustering mainly contains two steps according todata mining theory, data preprocessing and mining. Preprocessing of registeredproperties and realization of the algorithm are presented in detail in this paper.Additionally, two evaluation methods are put forward: one is evaluate the algorithm’sperformance in learner’s classification in the teaching system; the other is general test on IRIS data set, and compare the clustering result with FCMs. At the end, we cometo the conclusion: using the algorithm “Clustering based on Fuzzy EquivalenceRelation”to realize learner’s clustering is an effective proposal.