GLEIF-IT / lei-rdf

RDF LEI Project
MIT License
8 stars 4 forks source link

Bogus empty Geolocation created #207

Closed rivettp closed 3 years ago

rivettp commented 3 years ago

(as reported https://data.world/gleif/lei-data/discuss/remove-fake-geocoded-address-gleif-l1-datag/s5sxkfgr )

67597 entities are linked to gleif-L1-data:G- :

PREFIX gleif-L1-data: https://rdf.gleif.org/L1/ PREFIX gleif-geo: https://www.gleif.org/ontology/Geocoding/ select (count(*) as ?c) { ?adr gleif-geo:hasGeocodedAddress gleif-L1-data:G- } However, that geocoded address has no props:

select * { gleif-L1-data:G- ?p ?o } Examining the real addresses of these entities, we see plenty of data:

select * { ?ent gleif-geo:hasGeocodedAddress gleif-L1-data:G-; (gleif-L1:hasHeadquartersAddress|gleif-L1:hasLegalAddress) ?adr. ?adr ?p ?o } limit 100 These addresses are in Kuwait, Plovdiv, PLACE D'ARMES (mistake: that's not a city), Lagos, some Chinese city, etc.

So they look like real addresses that could not be geocoded.

In such case the link gleif-geo:hasGeocodedAddress should be omitted, rather than linking to the fake (empty) address gleif-L1-data:G-

rivettp commented 3 years ago

The XSL template already tests for gleif:geocoding_failed, will have to check the data to expand/replace that condition

rivettp commented 3 years ago

OK found the problem - the address was correctly being filtered but not the reference to it! Easy fix.