The dynamic queries by wireless mobile network users, will generate the social data with location tags and tracking data sequences, which enable the adversary can infer privacy information combined background knowledge, especially strong geographic correlation information. Therefore, we propose the Geo-matching privacy inference attack methods based on road network and sensitive semantic location. To address these issues, a k-implicit data publishing scheme with adaptive privacy budget is presented, which is based on the road network topological graph and the sensitivity quantification of grid unit, and also realizes an optimized dynamic anonymous region construction. Finally, the proposed Geo-matching attack algorithm is simulated to verify the effectiveness of the k-implicit data publishing scheme. The experiment results show that the proposed scheme can better resist the Geomatching attack under different privacy budget thresholds.