Hi Armin,
sorry I mis-read that.
If the descriptor is in the classpath, try
AnalysisEngineFactory.createAnalysisEngineDescription(String, Object...)
The string is the "package" and the name of the descriptor without ".xml" e.g.
AnalysisEngineFactory.createAnalysisEngineDescription("WhitespaceTokenizer")
AnalysisEngineFactory.createAnalysisEngineDescription("my.nice.package.WhitespaceTokenizer")
if the descriptor is on the file-system, try
AnalysisEngineFactory.createAnalysisEngineFromPath(String, Object…)
The additional parameters (Object…) are usually left empty, unless you need to
override parameters in the descriptor.
Regarding mixing TextMarker with your own pipelines, you may want to have a
look at the DKPro TextMarker stuff, if only for some inspiration how to do it.
[1]
-- Richard
[1]
https://dkpro-core-asl.googlecode.com/svn/de.tudarmstadt.ukp.dkpro.core-asl/trunk/de.tudarmstadt.ukp.dkpro.core.textmarker-asl
Am 30.04.2013 um 10:48 schrieb <[email protected]>:
> Hi Richard,
>
> I'm not talking about type system descriptors, but of analysis engine
> descriptors. I would like to create an AnalysisEngineDescription from an
> analysis engine descriptor file, e. g. like one the Textmarker Workbench
> created in a Textmarker Eclipse project. I'd like to add this
> AnalysisEngineDescription to AggregateBuilder and run the aggregate with
> SimplePipeline.run(). So I can mix Textmarker ae with my own.
>
> Cheers,
> Armin
>
>
> -----Ursprüngliche Nachricht-----
> Von: Richard Eckart de Castilho [mailto:[email protected]]
> Gesendet: Montag, 29. April 2013 18:18
> An: [email protected]
> Betreff: Re: uimaFit way of creating an analysis description from an XML
> descriptor file
>
> Hi Armin,
>
> the equivalent to your code would be something like
>
> TypeSystemDescriptionFactory.createTypeSystemDescriptionFromPath("/some/path/AnalysisEngineDescriptorFile.xml")
>
> However, I'd usually prefer to store my type system descriptors as a resource
> in my projects, so that they end up in JARs and on the classpath. To import
> one type system in another, I use "import by name" instead of "import by
> location".
>
> Type systems accessible via the classpath can be made known to uimaFIT, so
> that you don't actually have to load them manually. Many of the uimaFIT
> factory methods, like AEF.createPrimitive() exist in a variant that doesn't
> take a type system. In these variants, the type system is loaded
> automatically by uimaFIT. For details see [1].
>
> Hope this helps.
>
> Cheers,
>
> -- Richard
>
> [1] https://code.google.com/p/uimafit/wiki/TypeDescriptorDetection
>
>
>
> Am 29.04.2013 um 17:04 schrieb <[email protected]>:
>
>> Hi!
>>
>> In classical UIMA you use following code to create an
>> AnalysisEngineDescription from an XML descriptor file.
>>
>> final Path descriptorFilePath = Paths.get("/some/path/",
>> "AnalysisEngineDescriptorFile.xml");
>> final XMLInputSource xmlInputSource = new
>> XMLInputSource(descriptorFilePath.toFile());
>> final AnalysisEngineDescription analysisEngineDescription =
>> UIMAFramework.getXMLParser().parseAnalysisEngineDescription(xmlInputSo
>> urce);
>>
>> Now you can use analysisEngineDescription with
>> org.uimafit.factory.AggregateBuilder to create an aggregate analysis engine
>> and run the pipeline with SimpePipeline.run().
>>
>> How is the uimaFit way to do this?
>>
>> Cheers,
>>
>> Armin
>>
>>
>
>
