Usage of annotator "capabilities"

[Anton Shuster <br...@gmail.com>]

Hello all,

There is a capabilities section in the descriptor for an annotator where one
defines which annotations the annotator expects to find and which ones it
produces. However, despite how useful this information is, I can't seem to
find how it gets used (other than for documentation purposes).

In particular, I was wondering if there's some sort of verifier or something
that can (or does) check whether an that produces annotations of type A has
run before an annotator that expects annotations of type A. Also, it would
be useful if there was a flow controller that used the capability
information to call annotators in an order such that these dependencies are
satisfied (like in the case just mentioned).

Is there anything built into UIMA that uses this stuff?

Thanks!
--Anton

Re: Usage of annotator "capabilities"

[Marshall Schor <ms...@schor.com>]

Anton Shuster wrote:
> Hello all,
>
> There is a capabilities section in the descriptor for an annotator where one
> defines which annotations the annotator expects to find and which ones it
> produces. However, despite how useful this information is, I can't seem to
> find how it gets used (other than for documentation purposes).
>
> In particular, I was wondering if there's some sort of verifier or something
> that can (or does) check whether an that produces annotations of type A has
> run before an annotator that expects annotations of type A. Also, it would
> be useful if there was a flow controller that used the capability
> information to call annotators in an order such that these dependencies are
> satisfied (like in the case just mentioned).
>
> Is there anything built into UIMA that uses this stuff?
>
>   

Hi Anton,

Here are a few places where the capabilities are used.

There is a built-in flow controller, called Capability Language flow,
that uses this info - see
http://incubator.apache.org/uima/downloads/releaseDocs/2.2.2-incubating/docs/html/references/references.html
and search for Capability Language.

The Document Analyzer uses this information to decide what annotations
to show when displaying results.

The framework itself uses this information in setting up the default
"results specification" for a component, although the logic is fairly
complex - see the discussion here: 
http://incubator.apache.org/uima/downloads/releaseDocs/2.2.2-incubating/docs/html/tutorials_and_users_guides/tutorials_and_users_guides.html#ugr.tug.aae.result_specification_setting

The Component Descriptor Editor has a search capability for finding
annotators to add to an aggregate, that can search on types that are
declared to be output by candidate annotators.

-Marshall
> Thanks!
> --Anton
>
>