2002, Kavouras, M. and Kokla, M., "Developing Multi-scale, Multi-context Databases through the Semantic Integration of Heterogeneous Datasets" Print


Full title:


Kavouras, M. and Kokla, M., "Developing Multi-scale, Multi-context Databases through the Semantic Integration of Heterogeneous Datasets",  Proceedings of the 8th EC-GI&GIS Workshop, ESDI - A Work in Progress, 3-5 July, Dublin, Ireland, 2002. 




Nowadays, users’ growing demand to eliminate data duplication and avoid the timeconsuming and expensive process of data acquisition, as well as the increase in the availability of digital data have stressed the need to share and reuse existing data from different sources. Interoperability aims at the development of mechanisms to resolve any incompatibility and heterogeneity and ensure data exchange and reuse. Interoperability between different data sets requires the technical support for the exchange of data, but also the preservation of their underlying semantics. However, while technical issues related to data exchange have been developed sufficiently due to advances in information technology and standardization efforts (protocols, formats, interfaces), issues concerning semantics require further research.


The difficulty of geospatial data exchange between different sources lies, not so much in the incompatibility of data formats, but in the heterogeneity of data models, conceptual schemata and semantics. The achievement of interoperability between existing data sets is further complicated by the fact that this data are categorized according to different data standards and categorization schemata. Indeed, different interpretations of spatial data encoded in different data collections in conjunction with the complexity of spatial data are the main causes of semantic heterogeneity.


Semantic heterogeneity is caused by differences in the meaning, interpretation or use of the same or related data. Therefore, semantic heterogeneities arise due to different scientific fields – application contexts, varying levels of detail and time reference. For example, a road network as defined by a roadside land use project has a different semantic description from a road maintenance project with transport data.


At the European level different and diverse GI projects and data collections exist, e.g., for land cover – land use (CORINE Land Cover, CLUSTERS, PELCOM, LANES), ecology and environmental changes (ECOMONT, MEDALUS), agriculture (MARS, CLIVARA), forests (EFI), water (EU-Water Framework Directive), transportation (TEN). Projects with overlapping interests categorize their data according to diverse nomenclatures, a fact that hinders data exchange and association. It is difficult to convince for the imposition of harmonization standards on European countries, when such harmonisation has not been achieved or even pursued among the various EU GI projects. For example, CORINE Land Cover, CLUSTERS (Classification for Land Use Statistics: Eurostat Remote Sensing Project) and PELCOM are three European projects with similar objectives and interests: land cover – land use mapping. However, they use different land cover – land use categorizations.


Until recently, empirical approaches were used to resolve heterogeneities, in order to facilitate information exchange and reuse. However, empirical approaches only deal with specific problems and cannot provide an overall and uniform solution. Consequently, the development of specific, formal methodologies for resolving semantic heterogeneities and associating different data standards and categorization schemata constitutes an essential step for data exchange at the semantic level.


Download: here


[ Back ]





Do you like our new website?