Study On Classification Of E-commerce Users Cognitive Based On Neural Network

In recent years, the "user-centric" has become an service concept of many online services enterprise, and differentiated services reflects this philosophy. In the Past service is based on the demographic characteristics of traditional differences (such as gender, age, level of education, property, income, etc) to categorize users, but this kind of classification does not meet the needs of users, we should consider from cognitive dimension which is more closely linked with the user cognitive and psychological characteristics to classify user. Related studies have shown that cognitive differences will affect the behavior of the user’s network.cognitive theoretical studies have shown that cognitive level difference is dominant manner to measure by psychological test or scale methods, but have not yet been realized online hidden measurement. In this context, this paper proposes the use of neural network,a kind of machine learning classification methods, to achieve implicit measurement of the user’s cognitive types, Eventual provide differentiated services based on cognitive level.In this paper,firstly summarized the representation theory of the cognitive, include the need for cognition, cognitive style and cognitive schemata three cognitive dimension,from the meaning, the difference, the methods of measurement and the effect of online behavior.all these theory provide a theoretical basis for that the classification.Next,this paper summarizes relevant classification methods in machine learning,and select neural networks as a classification method, and comb in the e-commerce application classification method based on that classification recommended. all these theory provide a method basis for that the classification. Finally, we complete knowledge recommended experiment, and collect online behavioral characteristics of the cognitive differences.based on this sample data, we build user cognitive type automatic classification model using neural network classification method on the spss Clementine.All these work provide suggestions for differentiated service for recommendations based cognitive types.