I like the idea of using the Wiki and linking the type listed in a pipeline 
diagram to a type hierarchy tree page. Maybe to javadocs as a start?

I can pitch in on such documentation, but would not have the time to drive such 
an effort

From: N M [mailto:[email protected]]
Sent: Monday, May 12, 2014 9:07 AM
To: [email protected]
Subject: Re: cTAKES API documentation

a pdf is probably not the best format to contain this information, since 
updates are a hassle. I think this could be something that would be useful on 
the wiki, as versioning and history are much easier to control. I don't think 
it would be unreasonable to propose a working group to create something like 
this?
@James: thanks for providing the slides. they are a different "view" of the 
types, and important as well... I wonder if there would be some way to 
cross-reference a pipeline diagram (such as in the pdf) with a type inheritance 
diagram? One way I could see this working in a wiki setting would be to link 
the name of a type on a pipeline diagram page to its location on a type 
hierarchy tree page.
@Tim:
2) this depends on the pipeline. Some of the newer pipelines are slightly 
better at this, though even then the information is somewhat incomplete. Older 
pipelines (for example smoking) seem to be missing this information completely.
3) you're right that any UIMAfit components have the type annotations 
specified. Those cases really helped when I was creating my notes. It didn't 
seem to me that outputs were specified. it would be nice if a UIMAfit 
library/parser existed that could draw out this information. Though incomplete, 
it would help create a "fill-in-the-blank" initial structure, which could then 
be manually amended.
Is there any overall interest in making documentation like this more complete 
and available? I don't have the resources to do all of this myself, but if 
there are others who would like to pitch in, I'd gladly take part in the 
project.
-NM

On Wed, May 7, 2014 at 4:00 PM, Masanz, James J. 
<[email protected]<mailto:[email protected]>> wrote:
The slides I was referring to are:
https://wiki.nci.nih.gov/download/attachments/63996717/CTAKES_type_system_v1.0.5.ppt

The slides took the approach of highlighting the type system, not the 
annotators, showing which types of annotations were created and attributes 
filled in by each annotator, but without explicitly naming the annotator in the 
slide

From: Masanz, James J.
Sent: Wednesday, May 07, 2014 2:47 PM
To: '[email protected]<mailto:[email protected]>'
Subject: RE: cTAKES API documentation

There were slides created for the very first release of cTAKES that showed 
something along these lines. I don’t see it online anymore, but it’s outdated 
anyway. I’ll look for a copy and post them somewhere.

I like the PDF but I’d suggest keeping the format as simple as possible to make 
it easy to make updates.

Also, I suggest either removing or adding a footnote for those attributes that 
aren’t set by a given annotator.  For example, the output of Tokenizer 
Annotator is shown to be BaseToken, but because some attributes of BaseToken 
are listed, someone might wrongly infer that the Tokenizer sets the values of 
those attributes, when some of them are actually set by later annotators.

It would also be good to list which release of cTAKES it applies to.

-- James

From: N M [mailto:[email protected]]
Sent: Tuesday, May 06, 2014 4:01 PM
To: [email protected]<mailto:[email protected]>

Subject: Re: cTAKES API documentation

Hi,
Documentation which I was hoping to find as a user (hopefully eventually 
developer) of cTAKES, was a detailed description of each component (or analysis 
engine) and their interplay. As I've been looking at the cTAKES source code, 
I've been creating a chart for each pipeline that I've analyzed, hoping to 
essentially create a more useful component library than the one available at 
https://cwiki.apache.org/confluence/display/CTAKES/cTAKES+3.0+Component+Use+Guide
 (a "developer version" of this, if you will).
I've attached a pdf example of one of my personal note charts I've been working 
on, this one is specific to the UMLS plaintext aggregate AE. My idea is to have 
some sort of diagram for each pipeline that I can refer to, so that when I need 
to customize things in the future, I have a pretty good idea of where to go. I 
think it is important to chart inputs and output types especially.
Has anyone written a UIMA library to automatically document these kinds of 
things for UIMA components? running something akin to one would save a lot of 
time and effort, and would be much more maintainable than manually created 
charts. (of course, the inputs and outputs would have to be noted in the xml 
files, which doesn't seem to be the case for many of the AE's).

Looking at the previous discussions (notably the one regarding lvg entries), it 
seems to me that having a chart or other type of documentation which lists what 
types come out of and go into each component would go a long ways.
Does anyone know if anything similar exists?

Thanks.


On Fri, May 2, 2014 at 2:17 PM, ravi garg 
<[email protected]<mailto:[email protected]>> wrote:
Hi Xiayuan,
I hope this [1]  will be of help to you.

[1] 
https://github.com/ravigarg27/phenotips/blob/master/components/ctakes-service/src/main/java/edu/toronto/cs/phenotips/ctakes/AbstractScriptService.java
Regards,
Ravi

On Fri, May 2, 2014 at 10:07 PM, Miller, Timothy 
<[email protected]<mailto:[email protected]>>
 wrote:
James posted the links to the API docs earlier, but I suspect you may
find examples more helpful.

If you look at the classes under
/ctakes-clinical-pipeline/org.apache.ctakes.clinicalpipeline.runtime,
you will see how to instantiate the default pipelines from within
UimaFit using the xml descriptors.  You can override
BagOfAnnotationsGenerator with your own extractInformation() method and
do whatever you like for each annotation, or you can just copy the
layout to do something you like.

Alternatively, if you want to build your own pipelines programatically,
I would look at some of the pipelines in the ctakes-temporal project:
/ctakes-temporal/src/main/java/org/apache/ctakes/temporal/pipelines/TemporalExtractionPipeline_ImplBase.java
if you aren't up to date on trunk (i.e. you are working off of a
release), then look at the most recent version in svn:
http://svn.apache.org/viewvc/ctakes/trunk/ctakes-temporal/src/main/java/org/apache/ctakes/temporal/pipelines/TemporalExtractionPipeline_ImplBase.java?view=markup

Finally, there is work-in-progress to define factory methods for getting
commonly used pipelines, not ready for primetime but may give you some
more examples of how pipelines can be put together:
http://svn.apache.org/viewvc/ctakes/trunk/ctakes-clinical-pipeline/src/main/java/org/apache/ctakes/clinicalpipeline/ClinicalPipelineFactory.java?view=markup

Hope this helps.

Tim

On 05/02/2014 12:16 PM, Xiayuan Huang wrote:
> Actually I am looking for some API documentation like how to use and
> code with ctakes pipelines. Can anyone share the experience how do you
> write codes with some pipelines, (maybe in JAVA or else).
>
> Thanks,
> Xiayuan
>
>
> On 2014-05-02 10:55, Miller, Timothy wrote:
>> Yes Pei, thanks, though now that I see it I'm not as sure that it
>> will be useful to Maha and Xiayuan.
>>  Tim
>>
>> On 05/02/2014 11:17 AM, Pei Chen wrote:
>>
>>> Tim,
>>> Do you mean:
>>>
>> https://svn.apache.org/repos/asf/ctakes/tags/ctakes-3.1.1/ctakes-type-system/src/main/resources/org/apache/ctakes/typesystem/types/TypeSystem.xml
>>> [3]
>>>
>>> ?
>>>
>>> On Fri, May 2, 2014 at 11:14 AM, Miller, Timothy
>>> <[email protected]<mailto:[email protected]>>
>>>  wrote:
>>>
>>>> Another resource that may be more useful for many use cases is the
>>>> type
>>>> system documentation. I seem to have lost that link -- can anyone
>>>> please
>>>> chime in with it?
>>>>
>>>> On 05/02/2014 11:05 AM, Masanz, James J. wrote:
>>>>> Try
>>>>>
>>>>> http://ctakes.apache.org/apidocs/3.1.1/ [1]
>>>>>
>>>>> -- James
>>>>>
>>>>> -----Original Message-----
>>>>> From: Xiayuan Huang 
>>>>> [mailto:[email protected]<mailto:[email protected]>]
>>>>> Sent: Thursday, May 01, 2014 7:41 PM
>>>>> To: [email protected]<mailto:[email protected]>
>>>>> Cc: Maha Zohbi
>>>>> Subject: Re: cTAKES API documentation
>>>>>
>>>>> Yes, same question. I have been looking for API or
>>>> documentation for
>>>>> long. Could some one help guide to it?
>>>>>
>>>>> Thanks,
>>>>> Xiayuan
>>>>>
>>>>> On 2014-05-01 18:53, Maha Zohbi wrote:
>>>>>> Hi,
>>>>>>
>>>>>> We are looking to developing an application that processes
>>>>>> unstructured medical data and feeds it into a decision support
>>>> system.
>>>>>> To do that we will use cTAKES. We downloaded the binary and
>>>> followed
>>>>>> the user guide. However, I couldn't find any API
>>>> documentation.
>>>>>> Can anyone point me to API documentation for cTAKES or a
>>>> samples
>>>>>> guide to use the API?
>>>>>>
>>>>>> Thanks
>>>> --
>>>> Tim Miller
>>>> Instructor
>>>> Boston Children's Hospital and Harvard Medical School
>>>> [email protected]<mailto:[email protected]>
>>>> 617-919-1223<tel:617-919-1223> [2]
>> --
>> Tim Miller
>> Instructor
>> Boston Children's Hospital and Harvard Medical School
>> [email protected]<mailto:[email protected]>
>> 617-919-1223<tel:617-919-1223>
>>
>>
>> Links:
>> ------
>> [1] http://ctakes.apache.org/apidocs/3.1.1/
>> [2] tel:617-919-1223<tel:617-919-1223>
>> [3]
>> https://svn.apache.org/repos/asf/ctakes/tags/ctakes-3.1.1/ctakes-type-system/src/main/resources/org/apache/ctakes/typesystem/types/TypeSystem.xml

--
Tim Miller
Instructor
Boston Children's Hospital and Harvard Medical School
[email protected]<mailto:[email protected]>
617-919-1223<tel:617-919-1223>


--
Ravi Garg
4th Year
MSc (hons) Biological Sciences
B.E (hons) Computer Science and Engineering
BITS Pilani KK Birla Goa Campus


Reply via email to