Abstract: Accurate calculation and analysis of the threshold range of perceived value of different user groups from a group point of view is helpful to identify the differences of user groups, improve the level of paid reading service, and expand digital revenue. Combined with the factors affecting the willingness to pay for reading, this paper uses the K-means algorithm to cluster the paid reading users to obtain the user group, and carries on the M5 model tree modeling to the clustered user group data to determine the threshold interval of the perceived value of different user groups and data verification.It is found that there are significant differences in perceived value among different groups of reading users. On the whole, the perceived value threshold of the reading group has two special boundaries of 0.722 and 0.396 to define the different ratio interval. According to the threshold range of perceived value ratio applicable to different reading groups, the group can be defined as developmental type, mature type and radical type. Among them, there is an intersection between the threshold range of the mature reading group and the developmental reading group, but the applicable threshold range of the radical reading group will exceed the limit of 0.722 to form a special group.The innovation of this paper lies in that combined with the influencing factors of willingness to pay for reading, the perceived value is quantitatively defined and the quantitative research method is used to measure the threshold range of perceived value of different user groups, which is an in-depth exploration of the group behavior of paying users.

Key words: Digital reading, Paid reading, Perceived value, User clustering, M5 model tree