Semantics of Geographical Space and Geographical Data

A cooperative research proposal between:
Institute for Geoinformation, Technical University of Vienna, Austria
(local coordinator: Prof. Andrew Frank)
Directorate for Earth Observation, National Institute for Space Research, Brazil
(local coordinator: Dr. Gilberto Câmara)
School of Information Science and Technology, Penn State University, USA
(local coordinator: Dr. Frederico Fonseca)

General Objectives

This document presents a research proposal for an initial 2 year period. This proposal reflects the common interest of the partners on research topics in geographical information science. The research partners share a conceptual view that the "tiers of ontology" model (Frank 2003) provides a useful basis for semantical characterization of geographical space and geographical data types, a belief that algebraic specifications can be a useful way of expressing concepts in a formal basis (Frank 1999), and that is it is important to design and build information system architectures that capture innovative conceptions of GIScience theory.

Topic 1. Semantics of Geographical Space
One of the main challenges for the development of spatial information theory is the formalization of the concepts of space and spatial relations. Currently, most spatial data structures and spatial analytical methods used in GIS embody the notion of space as a set of absolute locations in a Cartesian coordinate system, thus failing to incorporate spatial relations which are dependent on topological connections and fluxes between physical or virtual networks. Representation of relative space (the relation of a spatial object to other objects) can be achieved partially in GIS by using arc-node data structures. However, the models associated to arc-node data structures are usually completely unrelated to the analysis methods that use representations of absolute space, a separation that leads to a limited conception of space in geographical information system. Research on this topic will be concentrated on the formal characterization of the different conceptions of space and on the development of analytical techniques that can express the combined relationships between "spaces of places" and "spaces of networks".

Paper:
Modeling Spatial Relations by Generalized Proximity Matrices
Ana Paula Dutra de Aguiar, Gilberto Câmara, Antônio Miguel Vieira Monteiro, Ricardo Cartaxo Modesto de Souza. V Brazilian Symposium on Geoinformatics, Campos do Jordão, 2003.

Topic 2. Algebraic Characterization of Spatial Ontologies
Current approaches to ontological characterization have been concentrated on concrete proposals for tools and techniques for building ontologies, such as the W3C Semantic Web. As a result, crucial notions such as 'interoperability' lack a formal definition, thus allowing different interpretations. Further advances in the field require abstract specifications of spatial ontologies and spatial data types. These specifications will serve a basis for the design and implentation of geoinformation system architectures based on spatial ontologies. Research will be initially concentrated in three topics:

• Algebraic characterization of interoperability between geo-ontologies: the aim is to propose algebraic structures that support the concepts of interoperability and transformations between ontologies.
• Ontology of fields: The field model represents spatial data as a set of continuous distributions. The aim is to produce an algebraic characterization of the different types of space-time fields, and to provide a formal definition for the operations on fields.
• Ontological characterization of map algebra: Map algebra is an established technique for processing raster data, which involves a set of operations that deal directly with the numerical values associated to each grid point. This situation gives rise to programs whose semantics are implicit and difficult to share and understand. The research work will characterize formally its operations in terms of spatial data types. The algebraic description of map algebra operations will allow the design of map algebra languages which are semantically meaningful.

Topic 3. Semantics of Geographical Data
Understanding how specialists make sense of the geographical data they obtain and exchange is an important contribution to the development of geo-ontologies. The current emphasis of semantical information processing is based on the concept that information needs to be structured and such structures should be standardized for achieving interoperability. This concept has led to proposals such as DAML and GML languages, that are not adequate for expressing the full range of relations and semantics which are present in different types of geographical data.
Our approach to the problem is to consider that geographical data has an inherent structure. Elements of this structure include data boundaries, data topology, and attribute names and their value distributions. This structure is the basis for the exchange of geographical data in real-life applications. This process of discovery of the semantics of geographical data starting from the data itself can be called "data wayfinding".
We will investigate techniques that would help a GIS specialist to obtain semantic understanding of spatial datasets, on a three-step process: (a) Conduct case studies of how specialists acquire knowledge about spatial datasets; (b) Formalize concepts and techniques used for "data wayfinding"; (c) Develop tools and techniques that help build geographical knowledge from spatial datasets.

Topic 4. Semantically-guided Geographical Information Systems Architectures
This research topic will investigate the use of semantics to build better geographical information systems. To achieve such goals, GIS architectures must support semantical concepts and have a modular design so they can be extended to include new research ideas(Fonseca, Egenhofer et al. 2002). Some specific areas of research include:

• Architectures for semantical information extraction in large image databases: our capacity to build sophisticated data collecting instruments is not matched by our means of producing information from these data sources. There are currently few techniques for image data mining in remote sensing archives, and thus we are failing to use the information available in our large earth observation data archives. We plan to conduct research on the design of information systems for remote sensing image databases that take into account an ontological characterization of images. This characterization considers that remotely sensed images are ontologically instruments for capturing landscape dynamics (Câmara, Egenhofer et al. 2001). This approach considers that geographic processes occur in a multi-scale space and result from the temporal and spatial interactions of different spatial phenomena over a physical landscape.

Generic GIS programming: There is limited theoretical and practical guidance on the overall problem of designing a GIS and establishing the interfaces between its various components. Since the abstractions used in GIS design are usually expressed as spatial data types, providing guidelines for GIS design requires a discussion of the paradigms used to define type systems that provide modularity and extensibility.
Our research will concentrate on the generic programming paradigm, which is based on the idea that there are fundamental laws that govern the behaviour of software components and designs interoperable modules based on these laws. This design allows spatial algorithms to be abstracted from spatial data structures, and GIS systems to be designed in a more modular and rigorous way.

References

Câmara, Gilberto, Max Egenhofer, Frederico Fonseca, and Antonio Miguel Monteiro. "What´s In An Image?" In Spatial Information Theory: Foundations of Geographic Information Science. International Conference, COSIT 2001., edited by Dan Montello, 474-87. Santa Barbara, CA.: Springer, 2001.

Fonseca, Frederico, Max Egenhofer, Peggy Agouris, and Gilberto Câmara. "Using Ontologies for Integrated Geographic Information Systems." Transactions in GIS 6, no. 3 (2002): 231-57.

Frank, Andrew. "One Step up the Abstraction Ladder: Combining Algebras - From Functional Pieces to a Whole." Paper presented at the COSIT - Conference on Spatial Information Theory 1999.

Frank, Andrew. "Ontology for Spatio-temporal Databases." In Spatio-Temporal Databases: The Chorochronos Approach, edited by Manoulis Koubarakis and Timos S