ctakes-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Andy McMurry <mcmurry.a...@gmail.com>
Subject Re: cTAKES - Context Annotation
Date Wed, 01 May 2013 10:15:10 GMT
RE: Context Dependent Annotation  

I had needs for fast local write (Consumer) that could directly be imported into a SQL database
for further analysis independent of cTakes. 
Consuming annotations are MUCH faster than what I was doing before 

Supports fast CSV->SQL inserts for the following types in a standard database table format

* Sentence (annotation) 
* BaseToken 
* NewLineToken 
* ContractionToken 
* PunctuationToken
* SymbolToken
* WordToken
* NumToken 

Creates database tables for each token type. 

If folks are interested in "yet another annotation consumer" I can upload.

--Andy 



On May 1, 2013, at 3:02 AM, "Masanz, James J." <Masanz.James@mayo.edu> wrote:

> I was thinking something like this:
> 
> public void process(JCas jcas) throws AnalysisEngineProcessException {
> for (Sentence sentence : JCasUtil.select(jcas, Sentence.class)) {
>  int sentenceBegin = sentence.getBegin();
>  int fromIndex = 0;
>  String s = sentence.getCoveredText();
>  while (fromIndex < s.length()) {
>    LookupWindowAnnotation lwa = new LookupWindowAnnotation(jcas);
>    lwa.setBegin(sentenceBegin + fromIndex);
>    int commaPosition = s.indexOf(COMMA, fromIndex);  
>    if (commaPosition > 0) {
>       lwa.setEnd(sentenceBegin+commaPosition+1); // include the comma 
>    } else {
>       lwa.setEnd(sentence.getEnd());
>    }
>    lwa.addToIndexes(jcas);
>    fromIndex = lwa.getEnd()-sentenceBegin;
>  }
> }
> )
> 
> -- James
> 
>> -----Original Message-----
>> From: dev-return-1566-Masanz.James=mayo.edu@ctakes.apache.org [mailto:dev-
>> return-1566-Masanz.James=mayo.edu@ctakes.apache.org] On Behalf Of Piyush
>> Jain
>> Sent: Tuesday, April 30, 2013 8:55 AM
>> To: dev@ctakes.apache.org
>> Subject: Re: cTAKES - Context Annotation
>> 
>> Hello James,
>> 
>> Thanks for your reply. Apologies for late reply.
>> 
>> Could you please help me on how can I create a new WindowAnnotationClass
>> for keeping comma out of scope of a sentence?
>> 
>> Regards,
>> Piyush
>> 
>> 
>> On Wed, Apr 3, 2013 at 11:14 AM, Masanz, James J.
>> <Masanz.James@mayo.edu>wrote:
>> 
>>> In addition to "MaxLeftScopeSize" and "MaxRightScopeSize", the context
>>> annotations created by ctakes-ne-contexts use the parameter called
>>> WindowAnnotationClass to control the boudaries.   By default, when the
>>> context annotator collects the context annotations for a named entity,
>>> it will not look beyond the boundaries of the sentence that the named
>>> entity is found in, because within NegationAnnotator.xml and within
>>> StatusAnnotator.xml, WindowAnnotationClass is set to be Sentence
>>> (org.apache.ctakes.typesystem.type.textspan.Sentence)
>>> 
>>> If you want a narrower range (to not cross commas) then I suggest you
>>> create new annotations whose boundaries are determined by the start
>>> and end of sentences and by occurrences of commas, and use the name of
>>> that new annotation as the WindowAnnotationClass.
>>> 
>>> -- James
>>> 
>>> 
>>>> -----Original Message-----
>>>> From: dev-return-1419-Masanz.James=mayo.edu@ctakes.apache.org [mailto:
>>> dev-
>>>> return-1419-Masanz.James=mayo.edu@ctakes.apache.org] On Behalf Of
>>>> Piyush Jain
>>>> Sent: Monday, April 01, 2013 10:56 AM
>>>> To: dev@ctakes.apache.org
>>>> Subject: cTAKES - Context Annotation
>>>> 
>>>> Hell Sir/Madam,
>>>> 
>>>> I am new to this forum and I have few question regarding cTAKES.
>>>> As of now, I am posting my question in this email only, please let
>>>> me
>>> know
>>>> if I have to post my questions somewhere else.
>>>> 
>>>> Question:
>>>> I am using cTAKES context annotation. I am trying to keep
>>>> "MaxLeftScopeSize" and "MaxRightScopeSize" as 2, earlier it was 10.
>>>> Also, I need to keep the context annotation look up to be limited to
>>>> same sentence and same "," span.
>>>> 
>>>> Could you please help me with this or if you are not the right
>>>> person could you please redirect me to right person?
>>>> 
>>>> 
>>>> Regards,
>>>> Piyush
>>> 


Mime
View raw message