Abstract:Aimed at analyzing users trajectory similarity on LBSN

the check-in points were hierarchical clustered

and the similarity of users’ track on each layer was calculated

then the geospatial similarity of the users’ behaviors was obtained

the Adaptive-Density-Clustering-Based User Trajectory Similarity Double Weighted Model was proposed. Specific to the distribution of users’ check-in points

the Clustering-Area-Radius-Based Adaptive Density Clustering Algorithm was proposed

the check-in points were clustered and the layered clustering areas which with certain area radius were obtained. The User Trajectory Similarity Double Weighted Model was put forward when calculating the users’ similarity: different weights of various levels and different weights of different areas in the same level. Then the similarity of users’ track on each layer was calculated based on the hierarchical clustering area. And due to different characterization ability for users’ similarity degree in the different areas

the similarity in specific level was calculated by the different weight. After that

the geospatial similarity of the users’ behaviors was obtained by overlapping the different scale user similarity value with different weighted. The simulation results demonstrate that this method can effectively analyze the users’ similarity of the track

and has higher accuracy.