Application of Collaborative Filtering Algorithm in Education Platform based on Pattern Recognition

In order to improve the accuracy of personalized recommendation technology, first give a definition for "empty state" in the state space of multidimensional semi Markov process, to obtain the extended multidimensional Markov process, which is combined with social network analysis theory to obtain social network information flow model. this model describes the process flow process of information between members in the network society. Then, based on social network information flow model, collaborative filtering algorithm SMRR (Semi-Markov and reward renewal) is put forward. Experimental results show that due to the comprehensive consideration of user's preferences and the effect of other members in the social network, the prediction accuracy of SMRR is obviously higher than that of the original algorithm.