Summary: | Many historical sources contain numerous names of places, the
manual assignment of which ties up a lot of resources. To simpliy this, an algorithm
is described with which such urbanonyms can be geocoded automatically. It is also
possible to cluster the places according to their common historical administrative
affiliation. Problems such as identical terms for place names are solved primarily
by including further information from the same context (same source). A validation
is done on the basis of about 3.4 million mostly German-language place names from
the genealogical database GEDBAS. In summary, about three out
of four relevant place names can be identified and localised. More than 90 percent
of the identified place names can be assigned to their historical province.
|