Francis Harvey
EPFL-IGEO-SIRS
CH-1015 Lausanne
francis.harvey@dgr.epfl.ch
 
Interoperability can be understood in a number of ways. In a minimal sense, even the capability to transfer data from one computer system to another without transformation loss can be identified as interoperability. In a broader sense, interoperability can be taken to suggest the ability of different applications to interact dynamically, facilitating the smooth interface of multiple information sources. This paper examines interoperability in this second sense, specifically the role of semantics in facilitating the exchange of information.

Various technical solutions are known and are being worked on to organize the basic technical infrastructure for exchanging data. OpenGIS documents, for example, describe standardized data formats, protocols, and transfer mechanisms that in the near future will permit a technically smooth exchange of data. This level of interoperability will change the way people interact with geographic information technologies. This technical foundation is crucial, but wide-spread success will require further methods to preserve geographic and attribute meaning.

While, on one hand, issues connected to the technical transfer of data are resolved, the sharing of information still requires examination of the underlying semantic issues (Kuhn, 1994). The exchange of information and its inherent meaning, although implicated in the discussions of data exchange mechanisms, does not find the same lucid technical solutions, nor discussion. In the OpenGIS abstract specification the authors acknowledge that information exchange between "World Views" requires application specific knowledge beyond the existing technical description of interoperability and the Geographic Information Community (Open GIS Consortium, 1996).

Clearly, there is a need to develop mechanisms to facilitate the exchange of information between different Information Communities. Semantics plays a crucial role in linking the different conceptual worlds. The research I present here investigates an approach for considering semantic issues in the design of interoperable GIS systems. By articulating the semantics of each conceptual world in a rigorous manner, the technical prowess of interoperability can be harnessed to help resolve deeper data sharing issues. This approach relies on the construction of 'portals' between the different realms. In particular, this work builds on a concept coming from ethnographic studies of science and technology, called boundary objects.

Boundary objects, in many ways analogous to boundary markers in geographic space, are points of reference for multiple actors who have different disciplinary, institutional, and/or social perspectives on phenomena. A boundary object, for example the wetland layer in a county GIS, connects multiple perspectives through its arbitration of differences, but, at the same time, maintains distinctions between the actors. Taking the example of a theoretical wetland layer in a county GIS, an attribute COUNTY_WET may be used for areas accepted by the county administrator as wetlands following the Cowardin schema, another attribute FEMA_WET may delineate wetlands adjudicated by the Corps of Engineers, and the attribute RIP_WET could be used by ecologists to designate environmentally sensitive wetlands protected by the county's riparian habitat preservation ordinance.

In such a case, the different mandates and legal frameworks of these three agencies necessitate the differentiation of unique wetland inventories. Perhaps COUNTY_WET overrides the others for general county purposes, but cases may well arise when it becomes necessary to reconcile the different inventories. At present, the only methods we have for performing this activity is the cumbersome process of negotiation. Because of existing contentions and political wrangling, these processes frequently end up in "data wars" (King & Kraemer, 1993).

Although trenchant institutional differences can often require solutions outsides the realm of GIS, the costs and long-term damage they incur is ultimately to no one's advantage. Instead of getting caught up in long-term struggles, a design method for interoperability that nips these problems in the bud and creates a stable foundation for stable solutions is obviously preferable.

Clearly, this requires overcoming semantic differences. This can be done in multiple ways. I will describe a method under consideration at the Canton Vaud (Switzerland) that builds on the boundary object concept for purposes of design. Through an iterative process of focused group meetings to lay out differences and commonalities, each actor's semantics are laid bare and the foundation built up for constructing means of transforming different models. Called 'portals', 'gates', or 'paths' they connect semantic differences and commonalities providing means of transformation. This broadening of the discussions surrounding design are the basis for constructing more robust data transfer mechanisms that preserve the semantic integrity of the transferred information. This modeling approach is suitable for formalization and integration with existing technical frameworks, applying specifications and outlining behavioral concepts for interoperability. Chains, actions, threads, entities are behavioral concepts for the formalization of the requisite morphisms. Instead of wearisome negotiations, this approach opens ways of rigorously describing and formalizing interoperability operations, thereby enriching the exchange of geographic information.

Clearly, this approach is limited in this form to 'closed-settings.' These are administrative or corporate environments with a distinct number of actors and distinct purposes. 'Open distributed processing', such as the provision and exchange of global-change data for research, with highly variable uses, will require more advanced solutions that can build on this preliminary research.

In summary, the approach I describe here suggests a solution to enriching interoperability by bringing fundamental semantic concerns into the primary design process. Beyond the technical differences that interoperability has addressed so successfully, the broadening of considerations to include the semantics of interoperability is a crucial next step.

REFERENCES

King, J. L., & Kraemer, K. L. (1993). Models, facts, and the policy process: The political ecology of estimated truth. In M. Goodchild, B. O. Parks, & L. T. Stayaert (Eds.), Environmental Modeling with GIS (pp. 353-360). New York: Oxford University Press.

Kuhn, W. (1994). Defining semantics for spatial data transfers. In T. C. Waugh & R. G. Healey (Ed.), Sixth International Symposium on Spatial Data Handling (SDH '94), 2 (pp. 973-987). Edinburgh, Scotland, UK: IGU/AGI.

Open GIS Consortium (1996). The OpenGIS Abstract Specification: An Object Model for Interoperable Geoprocessing, Revision 1 (Project Document No. 96-015R1). Open GIS Consortium.