The goal of interoperating GISs is to achieve an automated process that will allow to use data and software services across the boundaries that their collectors and designers envisioned. An ad hoc integration may work in a few specialized cases, but it is often difficult or impossible to generalize such an approach. The deficiencies of ad hoc interoperability becomes apparent when interoperating subparts are supposed to be extended by integrating new components or adding functionalities or replacing subparts with new software pieces. The difficulties are primarily in the semantics of the diverse implementations. Compatible semantics of geospatial information is a key characteristic of interoperating GISs, and powerful methods to capture and describe geospatial semantics are critical.
Like for many other things, a solid theoretical foundation is expected to assist in the development of interoperating geographic information systems by providing rationales and frameworks and giving guidance for implementations. It is not the first time that "theory" is discussed within the realm of GIS. In the early days of GI Science, Frank called for a unifying spatial theory. Also, the NCGIA solicitation included a statement on the need for research towards a theory of spatial relations, a particular subset of a spatial theory. With the advent of the conference series on spatial information theory (COSIT), researchers found a forum to discuss various theory components. Of particular interest within that community is the linkage to cognitive aspects, how people think about geographic space, and the concept of Naive Geography, a core of this perspective on spatial information theory, was defined as a set of theories Òthat provide the basis for designing future Geographic Information Systems that follow human intuition and are, therefore, easily accessible to a large range of users.Ó With interoperability, a new light is shed on spatial information theory, with a focus on capturing semantics and making software services work together.
An important consideration for research is that interoperating GISs need to go beyond providing interoperability for different geometries. Geometry is an integral part of GIS interoperability, however, the formation of particular geometries through shapes, orientations, and connectivities is driven in the first instance by different ways to conceptualize the world. If one considers the differences in spatial data sets that users want to integrate, one quickly realizes that they stem from different conceptualizations more often than from the choice of different geometric data models. These conceptual issues are a significant part of the semantics of spatial information. We need better models that capture how people express meaning. Computerized information systems contain symbols that need people's interpretations. Successful communication means that sender (often the data collector) and recipient (often the user) have the same mental models. In the realm of interoperability, there are two (or more) senders and the recipient is charged with the task of understanding the intended meanings of the senders and to integrate that information.