【Purpose/significance】 To establish an evaluation index system of social reading user behavior data, analyze and evaluate

the representation data of user reading behavior and social behavior, and put forward suggestions for the social reading platform or APP to classify the development of users and stabilize user groups.【Method/process】 An evaluation index system was established from the two dimensions of reading and social interaction. The behavior data of Weread_Tencent were collected as research objects,and CPM was used to analyze and evaluate the sample data.【Result/conclusion】 Reading time is the most important indicator in the dimension of reading behavior. In the dimension of social behavior, the weight value of fans' likes and fans' attention is slightly higher than that of users' comments. Platforms or apps should attach importance to the influence of users. Rank the Weread_Tencent users ac⁃cording to the comprehensive evaluation value, and divide the samples into four types of core users, social users, growth users and reading users according to the normalized data of reading and social behavior.【Innovation/limitation】 User behavior is complex and changeable, and the study of behavior characteristics need to be combined with static and dynamic behavior data for multidimensional analysis.