Usually, the territory of a country is divided into a number of areas that are administrated by a local government. Such areas are called - administrative units. The systems of such administrative division are hierarchical with several layers or levels of administration. Municipalities, regions, provinces are examples of administrative units of different levels. However, often it is difficult to compare structures of administrative division between countries because they use language-specific names for levels of administrative division.
Lets see what names are used to refer to the corresponding administrative levels in The Netherlands, Norway, Spain and Finland.
Linked Data brings semantic interoperability by the means of ontologies. Ontologies are collections of concepts and relations between them described in a formal way. WIth the help of ontologies it is possible for example to model systems of administration units and relate it to each other avoiding misinterpretation caused by language-specific notions. Therefore, lets compare structure of administrative systems in the Netherlands, Spain, Norway and Finland.
In the example above, we used the power of linked data to map local administrative system to a generic system of levels. This was done on the level of concepts or ontological level (similar to T-box in descriptive logic). Linked Data allows even more - interrelating data on instance level between data sets. For example, if two data sets contain information about the similar objects (e.g information about the same municipality) we can link them and enrich one description of the municipality with attributes coming from the other data set. The example below uses links between administrative units and corresponding objects in the DBpedia data base. In this case, DBpedia serves as a linking hub of the Web of Data providing access to other resources. Therefore, for example, we can traverse those links to retrieve the name of an administrative unit in another language. For example, lets find how to spell what the Dutch call “s-Gravenhage”.
People are used to refer to geographical features using names. Mountains, rivers, lakes, towns, villages and even single trees can have their own names. OpenEls project published more than 20 million place names (toponyms) and locations as Linked Data. Standardised means for describing meaning of place names enable semantic interoperability between national data sets. Seamless access to such rich data allows conducting interesting research. For example, linguists could potentially use this data to analyse spatial distribution of common toponym roots. The following example shows locations of place that have “holy” as a part of the name. Obviously, the root “holy” is spelled differently in different languages. The Linked Data technology makes it possible to formulate a single query that addresses national endpoints in their native languages. In the example the following spellings are used:
The SPARQL query language is a very powerful and flexible way to retrieve data based on graph patterns. However, SPARQL lacks functionality for a fuzzy search of literal values. In other words, if there is a need to find the location of Amsterdam SPARQL requires proper spelling of this name. It is not possible to retrieve the location of Amsterdam if there is a typo (e.g Asterdam) in the spelling of the name. Moreover, for lay users it is very difficult to master SPARQL. For these reasons, we created a gazetteer service enriched with Linked Data URIs for toponyms.