Richard Eckart de Castilho
unread,Sep 25, 2022, 5:19:40 PM9/25/22Sign in to reply to author
Sign in to forward
You do not have permission to delete messages in this group
Sign in to report message
Either email addresses are anonymous for this group or you need the view member email addresses permission to view the original message
to Imran Hassan, inception-users
Thanks for the screenshots and examples.
So, in the JSON data for example, you can see that a POS and a NER annotation are at the same location:
{
"%ID" : 317,
"%TYPE" : "webanno.custom.POSNER",
"@sofa" : 30,
"begin" : 207,
"end" : 222,
"Tag" : "NNP"
}
and
{
"%ID" : 517,
"%TYPE" : "webanno.custom.POSNER",
"@sofa" : 30,
"begin" : 207,
"end" : 222,
"Tag" : "PER"
}
So for me that looks normal. I am not sure what you expect.
That said, I would strongly recommend to use different layers for POS and NER because POS should be "per word" and NER can be one-or-more words. They do not have the same granularity.
Also, a POS layer should be configured to disallow stacking. A NER layer might allow stacking, but if it is not strictly necessary, I would also recommend to disable stacking for NER - in particular if you intend to use recommenders (because most do not support stacked annotations).
-- Richard