You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@uima.apache.org by "Thilo Goetz (JIRA)" <ui...@incubator.apache.org> on 2008/10/02 19:25:44 UTC
[jira] Closed: (UIMA-1193) Tagger throws occasional NPE
[ https://issues.apache.org/jira/browse/UIMA-1193?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Thilo Goetz closed UIMA-1193.
-----------------------------
Resolution: Fixed
Ok, I think I have it this time. I moved two static member variables into a method body, but don't think that was the issue.
The issue I think was that for some reason, sometimes an Ngram wasn't contained in the table of transition probabilities. So the code now checks for this and sets the transition probability to 0 when it happens. Eugenie, please take a look and see if you agree.
I also added a test case that runs the tagger on Moby Dick and compares the output against a gold standard, the gold standard being what the tagger produced before I fixed the issue. So I'm reasonably confident that I didn't change the semantics. Note: the test case only works if uimaj-core resides next to the tagger project, as I'm using the Moby Dick text from the test resources of that project. I was unable to figure out a way that would always work without either copying moby.txt to the tagger project or making it part of our distribution -- neither of which we want, I think. So now if the test case doesn't find the text, it prints a warning and moves on. Since nobody will see the warning, this is less than ideal. If anybody has any bright ideas, let me know.
> Tagger throws occasional NPE
> ----------------------------
>
> Key: UIMA-1193
> URL: https://issues.apache.org/jira/browse/UIMA-1193
> Project: UIMA
> Issue Type: Bug
> Components: Sandbox-Tagger
> Affects Versions: 2.2.2
> Reporter: Thilo Goetz
> Assignee: Thilo Goetz
> Fix For: 2.3S
>
>
> Tagger throws occasional NPE in Viterbi estimation.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
Re: [jira] Closed: (UIMA-1193) Tagger throws occasional NPE
Posted by Marshall Schor <ms...@schor.com>.
Re: test case / big file
I may be wrong, but I believe that copying in SVN doesn't really "copy"
in terms of space used up. You do have to do an SVN copy, not a
ordinary copy + commit, though.
-Marshall
Thilo Goetz (JIRA) wrote:
> [ https://issues.apache.org/jira/browse/UIMA-1193?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
>
> Thilo Goetz closed UIMA-1193.
> -----------------------------
>
> Resolution: Fixed
>
> Ok, I think I have it this time. I moved two static member variables into a method body, but don't think that was the issue.
>
> The issue I think was that for some reason, sometimes an Ngram wasn't contained in the table of transition probabilities. So the code now checks for this and sets the transition probability to 0 when it happens. Eugenie, please take a look and see if you agree.
>
> I also added a test case that runs the tagger on Moby Dick and compares the output against a gold standard, the gold standard being what the tagger produced before I fixed the issue. So I'm reasonably confident that I didn't change the semantics. Note: the test case only works if uimaj-core resides next to the tagger project, as I'm using the Moby Dick text from the test resources of that project. I was unable to figure out a way that would always work without either copying moby.txt to the tagger project or making it part of our distribution -- neither of which we want, I think. So now if the test case doesn't find the text, it prints a warning and moves on. Since nobody will see the warning, this is less than ideal. If anybody has any bright ideas, let me know.
>
>
>
>> Tagger throws occasional NPE
>> ----------------------------
>>
>> Key: UIMA-1193
>> URL: https://issues.apache.org/jira/browse/UIMA-1193
>> Project: UIMA
>> Issue Type: Bug
>> Components: Sandbox-Tagger
>> Affects Versions: 2.2.2
>> Reporter: Thilo Goetz
>> Assignee: Thilo Goetz
>> Fix For: 2.3S
>>
>>
>> Tagger throws occasional NPE in Viterbi estimation.
>>
>
>