[Purpose/significance] This paper explores the construction of reading pavilion in libraries from the perspective of reading promotion. Based on the analysis of big data, this paper analyzes the differences of reading preferences among users of reading pavilion in libraries and puts forward the strategies for reading promotion of reading pavilion in libraries.[Method/process] After defining reading pavilion as a new form of reading promotion, 105 496 pieces of reading data from 12 libraries were encoded, descriptive statistics and multiple linear regression model tests were conducted using SPSS 21.0, and qualitative research methods were used to explore the reading promotion strategies of reading pavilion.[Result/conclusion] The result of case analysis shows that there is a significant negative correlation between the length of reading works and the number of playback. There was a positive correlation between the number of views of the reading works and the number of plays of the audience. Classical literature material, leisure work material and theme education material have a significant positive effect on the number of playback; After viewing leisure segment works, audiences have the highest playback rate and education themed works have the highest standard deviation. Different library types have significant effects on the number of readings played. Therefore, this paper proposes that the reading promotion strategy for reading pavilion can be promoted from three aspects: the operator can review the online work classification and service optimization, the library can carry out a series of reading promotion activities and guide in-depth reading, the reading association and other organizations, and the reader can actively play the role of the reading promoter.