The importance of geospatial information has been widely recognized in current internet-based applications. The development of a National Spatial Data Infrastructure (NSDI) can take full advantage of the service-based technology to easily facilitate the sharing of geospatial data between different organizations. Although many geospatial web services have been successfully built, very few of them have taken the correct use of data in later applications into consideration. Consequently end users may naively misinterpret the data they acquired and make wrong decisions without noticing possible discrepancy. Especially for applications that depend on the real-time acquisition of geospatial data, the risk may tremendously increase with the increasing of the variety and volume of data. We argue that it is necessary to include the knowledge about data in the development of core geospatial web services of NSDI, such that the distributed data will be automatically possessed of built-in information to unambiguously identify itself and allow clients' applications to correctly analyze the correspondence or difference between heterogeneous data collected from various resources. Two strategies for developing such a service-based framework are proposed in this paper, the first is to analyze and identify the essential requirements for geospatial web services in NSDI, and the second is to design a standardized framework that ensures the necessary information can be correctly conveyed according to the types of services. The merits of NSDI come from the successful and transparent integrated use of geospatial data. However, only the "correct' use of data can really ensure the making of correct decision. Every participating organization therefore must understand its roles in NSDI and contribute necessary information for making the success of NSDI.