Abstract
For the past decade, the demand for accessing geospatial information has consistently increased in Korea due to the increasing practical application domains. To meet these local demands, many clearinghouses have been set up and are maintained by various sectors ranging from institutions to government bodies. Furthermore, the increasing numbers of users have expressed the need to integrate geospatial information for interoperable geographic information systems (GIS). However, this exchange of data among GIS users is a very complex issue. To solve this problem, we propose a set of ontology strategies between spatial and non-spatial data sources. This paper begins by discussing about the various issues related to the role of ontologies in the geodata reuse in distributed GIS and then present an architecture of ontology for retrieving distributed geospatial information.

1. Introduction
Since 1995, the so-called the National Geographic Information Systems (NGIS) project which is one of the government's initiatives to build a nation-wide spatial data infrastructure (NSDI) in Korea, is intended to construct, distribute, and share common spatial data that are very similar to the framework data (FGDC, 1997). Through the coordinated efforts of many federal to local and institutional level organizations including the National Geography Institute, a set of topographic maps of three different scales (i.e. 1 to 1,000, 1 to 5,000, and 1 to 25,000), several necessary thematic maps and ortho-imageries have been developed. As the amount of spatial data available increases, different geospatial information communities, public authorities as well as private institutions, recognize the importance of the World Wide Web as a medium to distribute their data. Thus, to meet these demands, agencies in charge of building up these data began setting up its own clearinghouses for the dissemination of data. From the users' point of view, however, multiple clearinghouses create great inconveniences in terms of searching and finding the dataset they want. These processes require mechanisms to share data via networks. The main obstacle in data sharing is the lack of semantic interoperability (Pundt et al., 2002).

Ontologies can provide a ''common basis'' for semantic mapping between information communities (Bishr et al., 1999). Therefore, in this paper we describe how ontologies can be used to overcome the problems related to data sharing, taking into account their semantics. This paper discusses issues related to the use of ontologies considering their semantics GIS. To overcome the problems related to data sharing, we discuss issues related to the role of ontologies in the geodata reuse in distributed GIS and propose a system of ontology-based tools for retrieving distributed geospatial information.

In Section 2, we review related literatures that have already examined the above-mentioned issues. In Section 3, we present the definitions of ontologies and the components that are used to describe them. In addition, we show how ontologies can play an important role in the geodata reuse and propose an architecture of ontology for retrieving distributed geospatial data. In Section 4, some conclusions are given and future works are described.

2. Literature Review
In this section, we review past works related to the use of ontology in connection with the development of information systems and GIS, with geospatial information sharing, and with ontology-related system architectures. The use of ontology in information system implementation was discussed by Guarino (1998), and that in GIS implementation by Frank (1997) and Smith & Mark (1998, 2001).

Smith and Mark (1998) reported on human subjects testing to derive ontology of 'folk' geographical categories. This paper first demonstrated that geographical objects and kinds were not just larger versions of the everyday objects and kinds previously studied in cognitive science.

Kuhn (2001) presented another approach to ontology design. He used natural language texts that describe human activities. However, he failed to justify the choices of referents in a particular domain, and thus, his ontology describes how the world works for certain agents. Furthermore, he pointed out the close relation of activities and domain knowledge. Transparency can only be obtained by observing texts, a method known from object-oriented modeling. Although object-oriented models cannot be validated, but Kuhn succeeded in providing executable formal specifications of an ontology.

Raubal (2001) modeled an agent-based system to simulate a wayfinding in built environments. He needed an ontological grounding of his model to guarantee that the agent's behavior is cognitively plausible; the grounding was obtained by interviews. He distinguished medium, substances, and surfaces in the real world (his ontological part of the model) and physical, socio-institutional, and mental affordances (his epistemological part of the model) (Winter, 2001).

Frank (2001) investigated the various levels of ontology, which allowed him to clarify different types of consistency constraints to geographical data. This paper makes the point that an ontological system is not simply an epistemological representation of reality but it must include also the subjective view of the universe. He described a five tiers model for ontology, which goes from Tier 0: human-independent reality to Tier 4: subjective knowledge.

The first two papers focused on ontology design, which is directly related to the development of systems, and the other papers dealt with applications.

Also, two international cases of clearinghouses in distributed GIS were reviewed and compared. For the international cases, Geospatial One-Stop Service (GOS) of Federal Geographic Data Committee (FGDC) and European Geo-Portal (Smits, 2002) were selected. Both systems were found to be similar in many facets. First of all, these systems deal with data in distributed servers and do not maintain them in an integrated manner at one particular server. In case of the GOS of FGDC, all of the data and their corresponding servers are scattered across the continent whereas the European Portal has data in different countries or different thematic servers. Secondly, both the systems rely on internet to connect data servers, portals, and users that they are a sort of web-based community outlet. One thing to note is that because communication infrastructures are inhomogeneous along the regions the level of service in terms of speed is inhomogeneous as well. Thirdly, the data that are provided are not in the same data models that are needed because standardization has been aroused. Fourthly, even though there are sets of metadata for identifying data there still exist mismatches in semantics that ontology issue must deal with. Lastly, they neither intend to re-organize dataset nor re-structure system architectures of resources. Rather, they pretend a higher level of portal connecting and summarizing these heterogeneous resources so as to provide a one-stop access point. In these systems, there are much functionality for user-friendly access and manipulation. Fig. 1 shows the various concepts related to GOS of FGDC.


Fig.1. Concepts of GOS of FGDC (from FGDC (2002))
Compared to these cases, some of the clearinghouses in Korea show a bit different model. First, these systems provide data produced or were consigned from outside data providers. Therefore, the systems afford faster access than that of the distributed system. Secondly, as e-commerce has become widespread users can download dataset as soon as transactions for payment are completed. Besides the foregoing, the other facets are very similar to those of international cases: internet connections, heterogeneous data model, semantic discrepancy, common dataset, metadata, and some toolsets for easier user interface. These systems are set up based on many research results done previously.

Based on such a review and understanding, the use of ontologies to build GIS applications can facilitate between data integration and prevent problems, such as inconsistency between ad hoc ontologies built into the system from occurring.

3. Architecture of Ontology for Retrieving Distributed GIS Databases
This section provides a brief review of the various concepts related to ontology. Here, we will demonstrate how ontologies can be play an important role in the geodata reuse and proposes architecture of ontology for retrieving distributed GIS.

3.1 Ontology
An ontology is an explicit specification of a conceptualization in a specific domain (Gruber, 2001). It can be defined as a collection of the standard terms used in a specific field. In addition, this ontology formally defines the relations among the terms for expressing knowledge effectively. That is, ontology represents the standard for applying in a wide scale to express, share and reuse knowledge provided in a web document. In the context of a GIS, such a language system can be a database or interaction language, carrying information about the parts of reality represented in the system. Generally, the ontology is divided into three classes; domain ontology, task ontology, generic ontology (Fig. 2). The domain ontology represents the specialized or generalized ontology about the domain; the task ontology represents specialized or generalized ontology about task; and the generic ontology represents a higher level of concepts that are common, and it is useful to reuse.

The ontology is similar to metadata in both form and property, but it can additionally represent reasoning function, express, extract, share and reuse of knowledge.


Fig.2. Top-level ontology (from Guarino (1998), Fonseca et al. (2000))
3.2 Architecture to Enable Ontology-Based Retrieval
In order to share ontology in a distributed environment a mechanism that allows others to use it and build more specialized ontologies is needed. The extended markup language XML allows us to create new markup tags and for the tags to contain both text data and other tags. The World Wide Web Consortium (W3C) has developed two standards, the resource description framework, RDF, and RDF schema (RDFS).

RDF defines a simple model for describing the interrelationships among Internet resources in terms of named properties and values. In many ways, RDF resembles the entity relationship diagram. RDFS provides a rich set of constructs to define and constrain the interpretations of vocabularies used in certain information communities.

The Web Ontology Language (OWL) is designed for use by applications that need to process the content of information instead of just presenting information to humans. OWL facilitates greater machine interpretability of Web content than that supported by XML, RDF, and RDFS by providing additional vocabulary along with a formal semantics.

In this study, we design the GIS meta-models using the RDF, RDFS, XML, and OWL. In addition, we propose the OWL for specification of meta-models and with which to build ontology through the systematic approach. The ontology is represented in data and processes meta-model. The meta-models are proposed in a layered architecture. Fig. 3, shows the various layers: OWL layer, PSL-ontology layer, GIS application layer, and XML Schema layer.


Fig.3. GIS Meta-model
First, we will need the various ontologies (Fig. 4). For each application schema there is an application ontology that is described with the shared vocabulary of the corresponding domain. These ontologies provide a formal description of the integrated process of a data source. Therefore, a user can find and access suitable information in any open and distributed environment of current Spatial Data Infrastructures. To implement the system, several standards have been proposed, including the feature catalogue description metadata section of the corresponding ISO 19115 documents and OGC (Open GIS Consortium)'s Catalog Services Specification.

4. Conclusions and Future Work
In this study, we proposed an approach and architecture for ontology-based retrieval of geospatial information that can solve existing problems of semantic heterogeneity. During the search process, the use of predefined ontology can help the user to reduce time needed to interact with the system by narrowing the search focus. In addition, a query interface can provide users with hints derived from the ontology to guide the user to use the correct terms. Also the system can accept users' queries which are based on their own semantics. Users can find and access suitable information in the open and distributed environments of current Spatial Data Infrastructures (SDI) and future SDI.

In this study, we proposed the OWL for the specification of the GIS meta-models and with which to build ontology - data and process parts - by adopting the systematic approach. With this, meta-models are proposed, which is composed of OWL layer, PSL-ontology layer, GIS application layer, and XML Schema layer. For each application schema, we have to provide one application ontology including the feature catalogue description metadata section of the corresponding ISO 19115 documents and OGC's Catalog Services Specification.

The proposed architectures are still in its conceptual level, but we believe it will be useful for building up a national system for a powerful and easy-to-access ontology-based system. In addition, the increased need for mobile and Digital Multimedia Broadcasting (DMB) services, interoperable tools make the ontology-based approach that much more valuable for GIS. In the near future, research work on GIS will be focused on ontology-related issues with concrete applications.


Fig.4. Conceptual Design of Ontology-based System
References

• Bishr, Y., Pundt, H., Kuhn,W., Radwan, M., 1999, Probing the concept of information communities - a first step towards a road of semantic interoperability, Kluwer Accdemic publishers, Boston, MA, pp. 55-71.
• FGDC, 1997, Framework Introduction and Guide, Federal Geographic Data Committee, Washington, DC.
• FGDC, 2002, Geospatial One Stop Project Design.
• Fonseca, F.T., Egenhofer, M. J., Davis, C. A., and Borges, K. A. V., 2002, Ontologies and knowledge sharing in urban GIS, Computer, Environment and Urban Systems, Vol. 24, pp. 232-251.
• Frank, A., 1997, Spatial ontology: a geographical point of view, Spatial and temporal reasoning, pp. 135-153, Dordrecht, Netherlands, Kluwer.
• Frank, A. U., 2001, The rationality of epistemology and the rationality of ontology, Rationality and Irrrationality, Proceedings of the 23rd International Ludwig Wittgenstein Symposium, Vol. 29, pp. 110-120.
• KRIHS (Korea Research Institute for Human Settlements), 1996, Strategies for Setting up a System for Distribution of Spatial Data. KRIHS.
• KRIHS (Korea Research Institute for Human Settlements), 2004, A study on involvement strategy of GSDI. KRIHS.
• Gruber, T., 1991, The role of common ontology in achieving sharable, reusable knowledge bases. In T. Gruber, Principles of knowledge representation and reasoning, pp. 601-602, Cambridge, MA: Morgan Kaufmann.
• Gruber, T., 2001, What is an ontology? http://www.-ksl.stanford.edu/kst/what-is-an-ontology.html
• Guarino, N., 1998, Formal ontology and information systems, Formal ontology in information systems, pp. 3-15, Amsterdam, Netherlands: IOS Press
• Kuhn, W., 2001, Ontologies in support of activities in geographical space, International Journal of Geographical Information Science, Vol. 15, No. 7, pp. 613-631.
• Pundt, H., Bishr, Y., 2002, Domain ontologies for data sharing - an example from environmental monitoring using field GIS, Computers & Geosciences, Vol. 28, pp. 95-102
• Raubal, M., 2001, Ontology and epistemology for agent-based wayfinding simulation, International Journal of Geographical Information Science, Vol. 15, No. 7, pp. 653-665.
• Smith, B., & Mark, D., 1998, Ontology and geographic kinds, In T. Peucker, & N. Chrisman, International Symposium on Spatial Data Handling, Vancouver, Cannada, pp. 308-320.
• Smith, B., Mark, D. M., 2001, Geographical categories: an ontological investigation, International Journal of Geographical Information Science, Vol. 15, No. 7, pp. 591-612
• Smits, P., 2002, Architecture and Standards, INSPIRE.
• Visser, U., Stuckenschmidt, H., Schuster, G., Vögele, T., 2002, Ontologies for geographic information processing, Computers & Geosciences, Vol. 28, pp. 103-117.
• Winter, S., 2001, Ontology: buzzword or paradigm shift in GI science?, International Journal of Geographical Information Science, Vol. 15, No. 7, pp. 587-590.