In order to improve the quality of personalized service of social networks and study the significance of data cycle recommendation algorithm, a social network user behavior data cycle recommendation algorithm based on artificial intelligence is proposed to solve the problems of low similarity calculation accuracy, low recall rate and long time consuming of traditional recommendation algorithm. Firstly, a scoring matrix is established based on user behavior, and Pearson correlation coefficient is used to calculate the similarity between scoring matrix and network behavior data. According to the similarity, cooperative filtering is used as the core to detect user data content in social networks that need to be recommended. Secondly, Top-N method is used to generate user neighborhood set. Finally, it is realized. Social network user behavior data content cycle recommendation.The experimental results show that the proposed algorithm has high accuracy of similarity calculation, the accuracy of recommended results of network user behavior data cycle can be as high as 97.2%, and the recall rate of recommended results is high and the time consuming is low, which improves the quality of personalized service of social networks.