Background

GeoCrossWalk is JISC funded middleware implementing a digital gazetteer service and server for the UK academic Higher and Further Education community. The rationale behind the service is that there is currently no unified entry point to assist in geographic searching within the existing academic network, as each information provider/service adopts different geographic coding conventions (some use postcodes, others placenames, some grid references etc.). GeoCrossWalk is designed to make geographic searching transparent by ’crosswalking' these different geographies as illustrated below.

Click on image to enlarge.

GeoCrossWalk is more than just a simple lookup facility however, as every geographic feature stored in the gazetteer has its detailed geometry stored with it (i.e. a city would be stored as a polygonal footprint (co-ordinate list), a river as a linear footprint etc.). Holding the geometry as an integral attribute of the feature enables complex spatial searching based on relationships between features, e.g. is feature A within a distance of feature B?; what features are contained within feature C?; what features does feature D intersect?

The ability to derive the relationships between features implicitly by geometric computation is significant and provides more accurate results than can be ascertained by simple lookups based on hierarchical thesauri methods, as is traditional in gazetteers.

Furthermore, geography in the UK is very complex and geographic boundaries in particular do not always nest; for example, postcode geography does not nest with electoral geography.

GeoCrossWalk obviates the problem of variable geographic naming by coding geographic features based on a persistent and consistent coding convention - national grid references.

Again, by use of the implicit relationships that can be inferred from their geometries, it is possible to 'crosswalk' places to: postcodes ...or electoral wards... or health authorities etc... As part of the work, the need for a geoparser was identified. That is, software that can take a document/resource that contains placenames and automatically identify their occurrence. Having identified a placename as such, the next logical step is to compare that against entries in the gazetteer which provides a means to access its 'alternate' geographies. In the Figure above, for example, 'Knowsley' could be resolved as parish code 'BX003' or grid reference 340900, 392300 - 347217, 397660.

This methodology provides a means to explicitly georeference (i.e. attach a grid reference to) implicitly georeferenced material (such as ’Knowsley’). The result is that more powerful geographical based search strategies can be applied, e.g. find me all images of places along the river Tweed.