Search code examples
wikidata

getting only instance of human entries in wikidata

Is there a way to get only higher confidence wikidata entries?

For example how do I tell that this is not an article about a person: https://www.wikidata.org/w/api.php?action=wbgetentities&ids=Q19069838&language=en&limit=1

The instanceof attribute (P31) links to human (Q5).

This article is clearly not about the human but the memo

Solution

  • That's the problem with data: sometimes it's wrong, and as it's expected not to be, there is no way to work around, other than

    • Fixing it (I did it for Q19069838)
    • Training a program to detect this kind of errors for you, but then, please use it to fix those errors ;) Machine learning is currently being used as a way to detect vandalism in Wikidata, I hope it will be further developed to address the kind of issue you encountered.