[
https://issues.apache.org/jira/browse/UIMA-6338?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17299842#comment-17299842
]
Michael Stenger commented on UIMA-6338:
---------------------------------------
Checking that the annotations in question also end within the window at least
comes more naturally to me, but I guess just starting therein isn't
unreasonable either. I don't mind that.
And as for my second point, I'd opt for changing the implementation to suit the
guide. I don't see why some annotations should be ignored just because they
start at the same position.
> Ruta: Inconsistency Interpreter vs. Guide on VOTE
> -------------------------------------------------
>
> Key: UIMA-6338
> URL: https://issues.apache.org/jira/browse/UIMA-6338
> Project: UIMA
> Issue Type: Proposal
> Components: Ruta
> Affects Versions: 2.8.1ruta, 3.0.1ruta
> Reporter: Michael Stenger
> Assignee: Peter Klügl
> Priority: Minor
> Fix For: 2.9.0ruta, 3.1.0ruta
>
>
> Hi again.
> This may be a bug, but I believe the following is more of a misleading
> formulation in the Ruta Guide:
> Comparing the explanation in the official Guide for version 2.8.1 on
> condition VOTE to its actual behaviour and the source code, there are two
> things that I find confusing/misleading:
> - The Guide says on the range within which annotations of these two types
> have to be located in to count in this condition: "annotations of the given
> two types within the window of the matched annotation". However, practically,
> all annotations of these types are considered that START (first basic token)
> within the matched range.
> - Furthermore, the Guide says something about counting annotations, which
> from my experience isn't quite true: As long as they start with the same
> basic token, any number of annotations of one type accounts for only one
> occurrence as far as VOTE is concerned.
> In case my observations are correct, I'd like to suggest changing the Guide
> to reflect that behaviour.
> I could provide an example if necessary, but I think this gets the point
> across just fine.
> Best regards,
> Michael
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
