search

rdmpage / glasgow-geoparser

Simple geoparsing using a gazetteer based on Wikidata and FlashText search
1 stars 0 forks source link

The word "Island" is often linked to Iceland #1

Open rdmpage opened 1 week ago

rdmpage commented 1 week ago

The word "Island" is often linked to Iceland by mistake. For example, "New Britain Island" is interpreted as "New Britain" Q59771 and "Island", and as "Island" is the name for Iceland Q3740828 in several languages (e.g., German). This will happen if the name we have in the data that goes into the trie does not include "Island" as a suffix.

rdmpage commented 1 week ago

A crude fix would be to delete the alternative name "Island", but perhaps a better solution would be to add "Island" as a prefix to those cases where this might cause a problem?