Geometrical integration within spatial data infrastructures

Producing and sharing spatial data is rapidly increasing in information and high-tech age. On one hand, quantity of producing is growing, and on the other hand, shape differences, ambiguities and compactibility-based problems may occur during the processes of access, tranformation, integration, conflation, etc. Natianal Spatial Data Infrastructure determines data rules within producing and consuming cycle according to previously determined standards and leads all the policies on this period according to needs. In order to minimize the efforts on integration process, data is required being produced with standards. To analyze, combine or re-produce spatial datasets, decision makers and researchers need datasets from different sources. The same entity may be represented by different featured objects in different datasets depending on different projection, scale, accuracy, target and time. Therefore, some problems depending on geometric, topologic and semantic differences during combining datasets may be encountered. These problems affect matching process, one of the most challanging process of integration, negatively. A matching process, in general, establishes links between objects from different datasets that represent the same entities. Links build bridges among objects by identifing and linking them and so make datasets more interoperable. Users may use these links during integration processes such as data transfering, updating, conflation. Scientists developed many algorithms to match spatial data from different sources automatically. However, since most of these algorithms are data-dependent, they may not succeed in matching process of the data from varied sources. In this study, two different road network datasets in an area produced by public and private organisations in Turkey were matched authomatically by some of the existing matching algorithms (MatchingPlugin, RoadMatcher ve optimization model). The achievements of the algorithms in native datasets were tested by comparing manuel matching results with the automated matching results. Results show that the tested algorithms could not reach enough matching number.