Abstract: 【Purpose/significance】For the current information user behavior feature mining methods can not integrate the data, and
can not accurately calculate the mean value of the data in the sliding window of the time series, resulting in the low accuracy of feature mining. Therefore, a method of mining information user behavior characteristics based on time characteristics is proposed.【Method/process】The average value of the data in the sliding window within the time series is calculated to obtain the initial sequence vector,and then the user behavior is divided into several equivalent time slices. Through sampling and statistics of various user groups, the qualitative status of user behavior is obtained. The average query frequency is taken as the standard to observe the user's query behav⁃ior characteristics and output the information mining results.【Result/conclusion】The experimental results show that the proposed method can mine more user behavior information at night than in the day, and more rest days than working days. In the case of network fluctuation, the time consumption is increased, but it is within a reasonable range. Compared with the traditional method, the proposed
method has lower mining error and strong applicability. The above experimental results prove that the research of information user be⁃havior characteristic mining based on time characteristics can obtain more accurate user behavior intention, improve the accuracy of user interest prediction and optimize the effect of network service.【Innovation/limitation】In order to further improve the efficiency of network information feature mining, we will focus on the parallel analysis of multiple network users' behavior in the future, so that this method is more suitable for massive network information processing.