The most significant feature of Location-based GIS is its service content may vary dependent on the situation, demand, and location of individual users, such that users may always get updated information as long as they can access to the internet. Though the bandwidth has been tremendously improved in the past few years, how to intelligently determine the appropriate service content remains to be a major challenge to most service providers. The service must have a better understanding about users' current situation and environment factors, so as to analyze possible data request from users and automatically adjust service content. Based on the location-aware concept, we proposed a server-side data selection framework capable of dynamically selecting and organizing corresponding data that meets users' demands based on their locations, request, and even personal information. The introduction of data selection knowledge allows on-the-fly data combination to generate service contents that does not exist beforehand. Furthermore, data mining technology can be applied to the users's profile to derive useful information for specifying more customized constraints to business database (e.g., interested goods on sale), such that the provided service content can be more precise and accurate.