I am currently using the MediaWiki API (through SPARQL SERVICE wikidata:mwapi) to query entities from Wikidata, using the wbsearchentities
endpoint.
However, I have noticed that the search results are not that good (example searching for charlton
or heston
, does not return Charlton Heston among the 10 first results, which would be expected), while the standard search endpoint query?list=search
works better.
My question is what algorithm is used for wbsearchentities
, and why doesn't it work as well as the standard query?list=search
?
My current understanding is that wbsearchentities
only searches labels, while query?list=search
is a full-text search, but it still should not justify this discrepancy in results IMO.
Cheers !