Identification Of Learners Based On Data Mining

Because of rapid development of Internet, various types of online learning systems are emerging, and on account of its convenient and user-friendly features, there is a growing trend that online learning system will replace traditional teaching method. It is necessary to identify the identity of learners study on line before given them result grade. But, it is hard to distinguish the authenticity of the identity of learners on the virtual network.Taking into account the combination of traditional methods such as user name and password authentication, it can only show the user having rights to log on, but could not explain the user as not pretending. Considering biometric technology such as face recognition authentication, it can identify the user, but this kind of technical is difficulty and popularized difficultly. When learners can not be monitored in real time, leaving only the circumstances of learners’ behavior properties, identification method based on data mining technology came into being.Data mining refers to analyze data, mining the underlying model through data. Identification of learners based on data mining means mining a series of behavior learners left to form learners’ behavior patterns, including normal and abnormal behavior patterns,and using proper clustering, classification and association rule algorithm to identify the identity of learners.The main contents are as follows:(1) We applied association rule algorithm of data mining to the identification of learners, forming into exceptional rule collection through association analysis, then we can identify the identity of learners by associated detection of their behavior, at last, we analyze the defect of this method.(2) We applied classification algorithm of data mining to the identification of learners, predicting the final grade authentic or not by classifying their final grade to identify the Identity of learners. We compare the result of identification using bayes, decision tree and neural network algorithm in order to select the best algorithm.(3) This paper brings up an idea that using bayes algorithm to assist decision tree algorithm, it makes up well for the defect of decision tree algorithm effectively, therefore, it makes identification by classification more effective.(4) This paper brings up an idea that using associated detection based on clustering algorithm to form more reasonable exceptional rule collection, it is more effective to detect learners’ behavior. Therefor, associated detection based on predicion by classification makes identification of learners more effective. This article is based on theoretical analysis of data mining and experimental verification, using three kinds of data mining algorithms which is classification, clustering and association rule to realize identification of learners, opening for the identification of new thought. Furthermore, we build the identity model based on personalized online learning system of the computer network course. It can effectively predict authenticity of the identity of learners and generalized to all kinds of online application system such as on line employee training system in enterprise and on line party system at school.