Abstract: Based on Web mining technologies, we examine the SDI user intensions including the background of domain users, access behaviors and the relationships of cartographic rules. A Web mining method is presented to construct the self\|adaptive domain space model and to analyze the user intension behaviors based on the service log documents focusing on the user preferences of spatial scale, feature semantic and feature object. By the analysis of the real service log information verifies the feasibility and effectiveness of this Web mining method.