Open the KIM GATE interface by running KIM/bin/kim gate
Create a "Large KB Gazetteer" resource with the following setup:
LKB Gazetteer set up
A brief description of the properties:
- annotationLimit - when the gazetteer creates the amount of Lookups indicated in this property value, it stops
- caseSensitive - whether the matching is case sensitive or not
- dictFeederClass - set to com.ontotext.kim.model.KimDictionaryFeederImpl
- dictFeederParams – set the directory you created for "Person"
- FeedSetupPath=$relpath$resources/gazetteer/dbpedia-person
- dynamicDictEnabled - set to false
- feedTransformerStages - additional transformations over the terms
- outputASName - the Lookup annotations are created in this set – "Person"
- relpath - set this to KIM/context/default/resources
- staticDictEnabled - set this to true
- staticDictSerializationPath - the cache is stored here
When the gazetteer initializes for the first time, it will look for a file named query.txt in the folder set in the dictFeederParams. The gazetteer will read the query from there and initialize its dictionary. Both SPARQL and SeRQL can be used. When you design your query, it is important to use the exact order and meaning of the query parameters - label, instance URI, direct class. The names are not important.
The outputASName is set to "Person". This means that the gazetteer will create its Lookup annotations in this annotation set. This is how you will differentiate between the recognized concepts by this gazetteer and other gazetteers.
Afterwards, save the application state to KIM/context/default/resources/IE.gapp.
If you want the gazetteer to create its dictionaries anew, you must remove the cache from the folder you have set in staticDictSerializationPath. In this case –
KIM/context/default/populated/gazetteer-person .