NCGIA Workshop on Interoperability for GIS

Portland, ME, August 10, 1995

Organizers: Max Egenhofer (University of Maine) and Michael Goodchild (UC Santa Barbara).

Participants

David Abel (CSIRO), Kate Beard (University of Maine), Kurt Buehler (OGC), Robin Fegeas (USGS), Kenn Gardels (UC Berkeley), Thanasis Hadzilacos (CTI, Patras), John Herring (Oracle Corp.), Kathleen Hornsby (University of Maine), Karen Kemp (UC Santa Barbara), Werner Kuhn (TU Vienna), G. Liu (ESRI), Scott Overmyer (University of Maine), Enrico Puppo (CNR), Glenn Smith (Lockheed Martin), Nectaria Tryfona (University of Maine), Andrej Vckovski (University of Zurich), Michael Worboys (University of Keele).

Summary

The goal of the workshop was to bring together scientists from the Open GIS Consortium (OGC) and database and domain experts to discuss the new directions and needs of an interoperable GIS environment.

The group argued that the central interest in this new direction of GIS is (1) to identify fundamental theoretical issues that focus on spatial data models and the way they relate to interoperating GISs, (2) to know what the users want to do with an OpenGIS, what they could do in that context that they can't do now, and (3) to find the fundamental impediments in interoperability and the way to overcome them. In particular, the participants came up with the first necessary steps to support the interoperability concept:

development of new methods (in addition to data exchange standards and current metadata approaches) for the better capture of the semantics and linguistic/cultural particularities of geographic information in order to publish, share and, integrate geographic data,
design of abstract, high-level spatial data and process models suitable for a wide range of application domains and,
definition of different alternative GIS architectures that would allow for open, distributed access to geographic information.

A set of metrics to evaluate operations on geographic data is also needed and a study of the properties of distributed operations on geographic data should be made. In addition, the development of benchmarks for the interoperability of current systems and data sets will be useful.

Furthermore, the examination of the way an OpenGIS architecture change the use of spatial information systems will give important results.

In order to discuss, further, specific topics, three groups were formed:

Group 1:
Lessons Learned from other Scientific Areas

Discussion Leader: Werner Kuhn, Technical University of Vienna

The topic of the discussion was to point out areas dealing with the issue of interoperability among systems and from which lessons can be learned. Fields like, standardization (e.g., SDTS, ISO, CEN), office automation, multimedia, human cooperation, human communication, operating systems, current technologies (WEB, distributed databases, COBRA, OLE, OpenDOC, computer graphics) and re-engineering (business processes, etc.) can give important results to the GIS community. Successful stories like, UNIX, Web, and telecommunications, as well as failures like, RISK must also be studied carefully.

Furthermore, specific concepts are framed in metaphors, such as the identification of metaphors/paradigms in each field, what do they achieve for interoperability, and how conflicts are resolved. Metaphors can also be used to: keep track of state, retrace history, deal with lifetimes, support sequencing, provide models of value, cost, charging, and reconcile inconsistencies.

Group 2:
Criteria of Success of "Good" Interoperable Models

Discussion Leader: Thanasis Hadzilacos, University of Patras

The goal was to identify research problems addressable by academic institutions. Issues like what are the hard questions to be answered and why and what are the questions that cannot be solved and are out of the goal of this effort should be discussed further .

On the other hand, criteria should be established for the successful data and process modeling in an interoperating GIS. A "good model," must capture meaning and linguistic/cultural differences, and allow mappings--which may be complete or partial, one way or two way--among various views on the same data, different understandings of the same data, different semantics of the same spatial information, similar semantics on differently structured spatial data, similar semantics on different representations of the same spatial entities/phenomena.

The model must be accompanied by an algebra of Geodata. Furthermore, a concept-concept mapping and analogies between the models that the remote systems use and the proposed model are critical.

Group 3:
User Scenarios and Prototype Definitions

Discussion Leader: Michael Goodchild, University of California, Santa Barbara

The topic of the discussion was to identify groups of users that will deal with new challenges and change the way they use an OpenGIS, and discuss user scenarios for each of these groups, i.e., scenarios where interoperability would allow users to do new things (e.g.. in terrain) or things they currently do and want to do easier. Three basic classes of users were examined:

users dealing with (geo)science: new algorithms must be used in order to manage data and new questions will appear because of the data fusion,
users dealing with decision making and policy: they will deal with semantic fusion (location, accuracy, thematic); there will be also the need of international collaboration and access to new techniques, and
users dealing with education and data mining: they will face metadata issues as well as questions like "what's the best data;" they also need better mechanisms for data browsing, access and retrieval.

General issues like, propagation of thematic error, positional and attribute accuracy, the need of a set of fusion tools, semantic fusion, are more or less common to all user groups and need to be considered in the new interoperable environment. Building prototypes will help to understand these needs as well as the change of the culture of the science that occurs.

Last updated on November 3, 1995.

[ Other NCGIA conferences | NCGIA-Maine | Other NCGIA sites ]
[ Department of Spatial Information Science and Engineering | University of Maine ]