You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@uima.apache.org by "Richard Eckart de Castilho (Jira)" <de...@uima.apache.org> on 2020/10/26 09:59:00 UTC

[jira] [Created] (UIMA-6286) select following finds zero-width annotation at reference end position

Richard Eckart de Castilho created UIMA-6286:
------------------------------------------------

             Summary: select following finds zero-width annotation at reference end position
                 Key: UIMA-6286
                 URL: https://issues.apache.org/jira/browse/UIMA-6286
             Project: UIMA
          Issue Type: Bug
          Components: Core Java Framework, uimaFIT
            Reporter: Richard Eckart de Castilho
            Assignee: Richard Eckart de Castilho
             Fix For: 3.2.0uimaFIT, 3.2.0SDK, 2.5.0uimaFIT


The uimaFIT {{selectFollowing}} as well as the UIMA v3 {{SelectFS.following}} show the following behavior:

* Given two annotations {{X: [0-1]}} and {{Y: [1-1]}}
* performing an operation {{select annotations FOLLOWING X}}, {{Y}} is returned.

So a zero-width annotation at the end of another annotation is treated to be "following" that other annotation.

However, a zero-width annotation at the beginning of another annotation is NOT treated to be "preceding" that other annotation.

* Given two annotations {{X: [1-2]}} and {{Y: [1-1]}}
* performing an operation {{select annotations PRECEDING Y}}, {{X}} is not returned.

Looking at the uimaFIT implementation, the fact that the SelectFS was modelled after the uimaFIT implementation and a general discussion on the mailing list regarding the rationales for interval predicates (e.g. following and preceding), I come to the conclusion that the uimaFIT selectFollowing was simply not thought to the end when it was implemented. So this seems to be an edge-case bug that needs to be fixed for a consistent behavior.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)