For example, can it distinguish between https://en.wikipedia.org/wiki/John_Howard and https://en.wikipedia.org/wiki/John_Howard_(Australian_actor)?
The entity linker performs syntactic similarity between a named entity mention in text (e.g.,
John Howard) and the label (e.g.,
foaf:name, …) and local name (e.g.,
John_Howard) of resources in a knowledge graph.
If that similarity is above a certain threshold, a link is created between the mention and the resource.
When more complex identification of entities is required, users are recommended to implement their own custom extractors.
So when we have two identical names that should be distinguished by context - say is it about acting or politics - Stardog doesn’t help?
Stardog does help, if you write a custom extractor that is able to distinguish context. Out of the box, we don’t provide an rdf-extractor that does that automagically, mostly because that distinction is always highly domain dependent.
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.