Search Mailing List Archives
[bioontology-support] Question Regarding BioPortal Annotation Service
frydawg524 at gmail.com
Mon Jun 3 20:58:02 PDT 2013
My name is Zach Fry and I am doing some research with the Tetherless World
Constellation at RPI. I am working on an Information Extraction Algorithm
that leverages the NCBO Gazetteer Ontology(Virtual ID 1397). I am working
toward entity disambiguation using the partonomic relationships defined in
the Gazetteer Ontology. I have a few questions regarding the results that
are returned using the NCBO Annotation Service and how the Annotator
Service recognizes candidate concepts.
When I search for "New York" with the Gazetteer Ontology, I do not get a
concept that returns the "State of New York", which would be the
correct concept that I am looking for. The annotator service only returns
concepts that have a preferred label of "York"
Similarly, when I search for the concept "Northern Highlands Lake District"
I do not get the correct concept "Northern Highland" but instead get
concepts "Highlands" and "Lake District".
I was wondering why I do not get the correct concepts? In the first case,
"New York" is a direct substring match to "State of New York", and in the
second case "Northern Highlands Lake District" is a substring match to
"Northern Highland", but a little off because "Northern Highlands Lake
District" has an extra "s" at the end of Highland and the added string
Is this an error from the Annotator Service, or is this because how the
lexical matching of a keyword to a candidate concept is performed? Would
there be a way to improve the matching to include this concept, or a
parameter that I could set that would return ALL possible substring
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the bioontology-support