Abstract: This paper proposed a new personalized recommendation method.The method is derived from the research on the essence of personalized recommendation technology.We arrived at a solution which includes several methods:one is calculating the offline similarity by the use of normal distribution’s convolution theories,one is calculating the offline similarity by calculating the number of similar operations between each pair of items,the last one is integrating the similarity results of different methods by the use of something similar to the Bayesian formula.We also mentioned some other methods and techniques used to engineering implementation.The method proposed in this paper has been successfully applied in the field of data mining.

Key words: Personalized recommendation,Similarity,Data mining