Abstract: To solve the problem of difficulty in obtaining personal privacy information of Internet users, incomplete and uncommon or even false content of the public information, a trajectory similarity analysis method was proposed based on time sliding window model. Firstly, some location data with universal, objective and reflecting capability for user behavioral habits were selected as similarity analysis base to collect the data of Sina Weibo and Didi Taxi, forming two high-value data sets with the characteristics of domestic netizens as experimental objects. And then, a trajectory similarity matching algorithm was developed based on the time sliding window. Adjusting the time window and the F value of the location distance optimization algorithm, it was arranged to realize the similarity analysis of users of different network platforms. Finally, some validation experiments were carried out based on the user location data of Sina Weibo and Didi Taxi. The results show that geographical location is a positive correlation factor for virtual identity similarity judgment, and the similarity average F-value can reach up to 90%.