gmane.comp.apache.uima.general

Ruta 2.0.2 Textruler problems

Sondes Bannour — 2013-05-23T11:49:14

Hi,

I tried the textruler view (whisk (token) ) under Ruta 2.0.2 and i got
weard results like rules with p=0 and n=0 while i fixed the maximum error
threshold to 0.1. Whisk (generic) doesn't even work for me (it seems like
crashing on the first step and still waiting without any results) (i am
just trying the pre-build version). I intend to do some development on
textruler but still don't find the appropriate version: i was using the
textruler source code with textmarker 1.0 but now that textmarker has been
improved i would like to use the latest version, the problem is that
textruler source code isn't working fine with these new versions of
textmarker.

cheers
Sondes

Ruta - Optional First Token Problem

2013-05-23T11:09:16

Hi!

In Ruta 2.0.2-SNAPSHOT, rules with an optional first element do not work. The optional part seems to be mandatory. Using

DECLARE Test;
"a"? "b" "c"{->MARK(Test, 1, 3)};

on

a b c x b c

marks "a b c" (0, 5) but not "b c" (8, 11).

Cheers,
Armin

Adding features to TokenAnnotation and DictTerm in Concept Mapper

Mansi verma — 2013-05-23T07:36:39

Hi

I am using UIMA for annotating some documents using Concept Mapper. I have
built the dictionaries and configured it to our requirements. However, I
wanted to add
 features to TokenAnnotation and DictTerm.

For Example existing TokenAnnotation annotation supports the following
features :
text, tokenType, tokenClass and uima.tt.tokenAnnotation. Now if I want to
add more features such as POS or group etc.

Is there a neat way of doing it without touching the conceptMapper.jar and
changing our typesystem to extend the two types ?


Thanks
Manisha

managing resources for UIMA?

Jens Grivolla — 2013-05-22T16:31:45

Hi, while not strictly a UIMA issue, we have a problem that seems very 
relevant in the context of UIMA analysis engines: how to manage large 
binary resources such as trained models used by an AE, etc.

So far, we have managed to achieve a good separation between code 
development and the actual AEs, using Maven (and git for version 
control). An AE thus consists only of a POM referencing the code, the AE 
descriptor, and the resources used for the AE. The AE poms are 
configured to generate PEAR archives that include all dependencies and 
resources.

At this point we have the code in git, and the AEs' pom and descriptor 
also, while we manually copy the resources to the directory before 
running `mvn package` (and exclude those resources from git). We're 
missing a way to manage those resources, including versioning etc.

I'm guessing that this is a rather typical problem, so what solutions do 
you use? We're thinking of having all resources also in Maven (e.g. 
Artifactory) so we can reference them with a unique identifier and 
version. This would also help us when moving to more complex pipeline 
assemblies using uimafit instead of generating individual PEARS for each 
component in order to create complete packages.

Btw, we are just very few core developers, with most of the team made up 
of linguists, so we want to make it easy for them to save versions of 
resources they create and assemble AEs by just referencing the algorithm 
and resource (e.g. "create a new OpenNLP POStagger using 
spanish-pos-model.bin, version 1.2.3").

Thanks for sharing your experiences with this...

Jens

Ruta - MARKFAST

2013-05-21T13:49:38

Hello!

Is there any possibility to match strings like

nC.
v. Chr.

with MARKFAST?

Cheers,
Armin

Ruta - Token Order

2013-05-21T10:47:43

Hi,

In Ruta 2.0.2-SNAPSHOT a token with begin offset 0 and end offset 2 comes before a token with begin offset 0 and end offset 0. The token order is not as I expected. Thus in my case, SourceDocumentAnnotation was the second token in the token sequence and the rule didn't match. It took me some time to find that out. The end offset of SourceDocumentAnnotation should better be the length of the text. How is the token ordering defined?

Cheers,
Armin

Ruta 2.0.2-SNAPSHOT - Eclipse Plugin Installation

2013-05-21T07:18:15

Hi!

I've checkout Ruta 2.0.2-SNAPSHOT with

svn checkout https://svn.apache.org/repos/asf/uima/sandbox/ruta/trunk

and build it succesfully with

mvn clean install.

Now, how to install the Eclipse plugins? Is there a local reposity or update site for Eclipse? Or, which files need to be copied manually?

Thanks,

Armin

Concept Mapper in code

harshal patni — 2013-05-21T06:49:06

Hello Everyone,
                   I am currently writing ConceptMapper in code (not using
XML files). Basically I am definitely AnalysisEngineDescriptions and
TypeSystem Description in java code. I create the following 2 parameters
using "ConfigurationParameter" class.

   1. AttributeList
   2. FeatureList

The problem is both these should be arrays. But ConfigurationParameter only
provide the  following types (String, Boolean, Integer, Float). When I pass
these values as string it fails and throws a java.lang.ClassCastException
error. COz it expects an array vs a String.  How should I solve this issue?
Below is the AnalysisEngineDescription code.
//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////

primitiveDesc = new AnalysisEngineDescription_impl();

 primitiveDesc.setPrimitive(true);

 primitiveDesc.getMetaData().setName("Concept Mapper Offset Tokenizer");

 primitiveDesc.setAnnotatorImplementationName(
"org.apache.uima.conceptMapper.ConceptMapper");



 ConfigurationParameter p1 = new ConfigurationParameter_impl();

 p1.setName("AttributeList");

 p1.setDescription("Attribute List");

 p1.setType(ConfigurationParameter.TYPE_STRING); // Here is the problem

 ConfigurationParameter p2 = new ConfigurationParameter_impl();

 p2.setName("FeatureList");

 p2.setDescription("Feature List");

 p2.setType(ConfigurationParameter.TYPE_STRING);


primitiveDesc.getMetaData().getConfigurationParameterDeclarations().setConfigurationParameters(
new ConfigurationParameter[] { p1,p2 });


primitiveDesc.getMetaData().getConfigurationParameterSettings().setParameterSettings(
new NameValuePair[] { new NameValuePair_impl("AttributeList", "canonical"),
new NameValuePair_impl("FeatureList", "DictCanon") });

       TypeSystemDescription typeSystem = new TypeSystemDescription_impl();

     dicTerm(typeSystem);

     conceptMapperTokenizer(typeSystem);

     baseTokenizer(typeSystem);

     primitiveDesc.getAnalysisEngineMetaData().setTypeSystem(typeSystem);

//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////

Also how should I pass the dictionary file in code?


Thanks,

Harshal

Aggregate Delegates and Remote Services

GATE User — 2013-05-20T20:56:00

I've created two primative AEs and tested them locally on my machine.  I then created a aggregate AE that calls each of the primitive AEs.  The aggregate AE was created using the Component Descriptor Editor plugin for eclipse and things went smoothly until I got to adding the Aggregate Delegates.  I clicked on remote, but I' then prompted to select either the SOAP or Vinci protocol.  The two primatives have been deployed and are running on a test AS system (2.4.1).  Which protocol should I be using, if either?

Since I wasn't sure which protocol to use, I made a copy of the AE descriptors for each of the primatives and added those using the "Add..." button.  Then, in the deployment descriptor for the aggregate AE, I set the delegates to run remotely and gave it the address of the broker and the queue name.  After creating the pear and deploying it, the aggregate AE deploys fine, but errors pop up when we try to run it.  The errors are:

      *** ERROR: line-number: 13 deployment descriptor for analysisEngine:  specifies async="true" but the analysis engine is a primitive

      *** ERROR: line-number: 13 deployment descriptor for analysisEngine:  specifies false for the async attribute, but contains a delegates element, which is not allowed in this case.

If the async="true" is removed, we get the following error:

      *** ERROR: line-number: 16 The delegate in the deployment descriptor with key="MgrsRegExQueue" does not match any delegates in the referenced descriptor

      *** ERROR: line-number: 25 The delegate in the deployment descriptor with key="MgrsValidatingAnnotatorQueue" does not match any delegates in the referenced descriptor

I've compared the deployment descriptor to the example one, and nothing is jumping out at me as wrong.  Is this a problem with the deployment descriptor or AE descriptor?  Thanks in advance.

Changing the original text based on annotations

GATE User — 2013-05-20T02:21:09

1)  How do I change the original message based on annotations in UIMA.  For example, lets say I have the string:
201301012345

That contains both the date and time.  I want to have an annotator that will find such things in the text and add a space between them so it becomes:
2030101 2345

What's the easiest way to modify the text in this instance?

Also, let's say I have the sentence:

See Spot run far down Main Street.

and I have an annotator that that finds and labels main street as a street name.  Now I want to make an annotator that, if it finds a street name annotation, to change that street name into something else, like River Blvd.  So the above sentence would be:

See Spot run far down River Blvd.

What's the easiest way to do this?  Will I, afterwards, have to resend the CAS through the pipeline again or is there an easy way to update all annotations that would be affected by the change since River Blvd is shorter than Main Street?

Thanks in advance.

Ruta 2.0.2 - Grouping Problems

2013-05-17T10:34:10

Hello!

Let A, B, C, D and F denote type names. Then, A B? C D{->MARK(F, 1, 4)} works.
A (B)? C D{->MARK(F, 1, 4)} causes a NullPointerException.
(A B)? C D{->MARK(F, 1, 4)} causes an ArrayIndexOutOfBoundException: -1.

Any ideas?

Cheers,
Armin

Ruta - RETAINTYPE

2013-05-16T12:21:35

Hello!

In Ruta SNAPSHOT-2.0.1 Document{->RETAINTYPE(ALL)}; does not retain SPACE. It is the same with ANY and WS.

By the way, where can I get the newest version of Ruta (jar, svn, etc)?

Cheers,
Armin

Processing multiple CASes from different AAEs together

Mateusz Parzonka — 2013-05-09T20:57:30

Hello,

what is the best practice to process CASes resulting from multiple and 
different AAEs? Assume I have an AAE1 processing document collection 1 
and an AAE2 processing document collection 2. Now I would like to 
compare each tuple of the resulting CASes of AAE1 and AAE2 using a third 
AAE. My underlying intention is to "compare" each document in collection 
1 with each document in collection 2 using different preprocessing 
pipelines.

Is there some abstraction in UIMA to perform this? I saw an solution 
performing an n-cross-m comparison using two views where each view 
represented a different document. This works but seems a bit inflexible, 
assuming you need to configure different processing pipelines for each 
collection type.

Regards,
Mateusz

Comparing each document from collection 1 with each document from collection 2

Mateusz Parzonka — 2013-05-10T12:59:36

Hello,

what is the best practice to process CASes resulting from multiple and 
different AAEs? Assume I have an AAE1 processing document collection 1 
and an AAE2 processing document collection 2. Now I would like to 
compare each tuple of the resulting CASes of AAE1 and AAE2 using a third 
AAE. Simply speaking, the intention is to "compare" each document in 
collection 1 with each document in collection 2 using different 
preprocessing pipelines.

Is there some abstraction in UIMA to perform this? I saw an solution 
performing an n-cross-m comparison using two views where each view was 
associated to a different document. This works but seems a bit 
inflexible, assuming you need to configure different processing 
pipelines for each collection type.

Regards,
Mateusz

CasCopier and derived types of DocumentAnnotations

Georg Fette — 2013-05-08T12:17:05

Hello,
I have a type system in which I have created a type which inherits from 
DocumentAnnotation. As DocumentAnnotation is not inheritanceFinal I 
thought that this is allowed. I have now a CAS in which I have several 
annotations of that derived type. In the method copyFSInner of CasCopier 
the proper target feature structure to copy instances of my type the 
first DocumentAnnotation in the index of documentAnnotation of the 
target CAS. Is DocumentAnnotation to be thought a singelton in a CAS ? 
Why isn't it prevented in the framework to derive from it or to create 
further instances of that type and add it to the CAS ?
If this is done the CasCopier can not be properly used.
Greetings
Georg

AnalysisEngineDescription XML

harshal patni — 2013-05-08T10:04:57

Hello Everyone,
                      I am trying to call AnalysisEngineDescription XML
files within UIMAFIT.  Here is what I do..

 CollectionReader reader =
UriCollectionReader.getCollectionReaderFromFiles(files);

XMLInputSource in1 = new XMLInputSource("desc/analysisenginedescription.xml"
);

     AnalysisEngineDescription aeDesc2 =
UIMAFramework.getXMLParser().parseAnalysisEngineDescription(in1);

AggregateBuilder builder = new AggregateBuilder();

builder.add(aeDesc2);

SimplePipeline.runPipeline(reader, builder.createAggregateDescription());

Is this right? I have some XML files I had built using Component Descriptor
earlier. I was using this within UIMA pipeline. I recently wanted to
integrate POS Tagger (ClearTK) within my pipeline. This when I cam across
UIMAFIT...

I keep getting an error whenever I run this code?

Thanks,

Harshal

Textmarker/Ruta - negative look behind/ahead

2013-05-08T06:42:39

Hello!

Is there a mechanism like negative look behind or negative look ahead in Ruta?

Cheers,
Armin

PhD Research Studentship, Natural Language Processing, University of Sheffield, UK

Genevieve M Gorrell — 2013-05-07T17:06:48

Applications are invited for a fully funded PhD studentship on computing
the veracity of social media content.

Application closing date is* 31 May 2013. *

The aim of this studentship is to design natural language processing
methods to compute the veracity of social media content and deal with the
specifics of medical language. The goal is to model, identify, and verify
healthcare-related misinformation and disinformation, as they spread across
online media (e.g. patient forums) and social networks. The studentship is
hosted at the Department of Computer Science at the University of
Sheffield, UK.

For more information, see:

http://www.jobs.ac.uk/job/AGL052/phd-research-studentship/

Same message sent twice to a single Analyze Engine when using Flow Controller

Spico Florin — 2013-05-07T09:36:22

Hello!
  I'm using UIMA AS 2.3.1 with ActiveMQ Broker 5.4.1. My pipeline has the
following structure:
2 Remote AAE that are running in parallel
1 Remote AE that should run serially after the above two has been executed
1 Parent AAE with flow controller that is controlling the above three AE
My flow controller configuration looks like this:
   <configurationParameterSettings>
      <nameValuePair>
        <name>Flow</name>
        <value>
          <array>
            <string>RemoteAAE1,RemoteAAE2</string>
            <string>RemoteAE3</string>
           </array>
        </value>
      </nameValuePair>
<configurationParameterSettings>
    My problem is that at some point, a message that was sent ONCE to the
pipeline is processed TWICE by the RemoteAE3 and my scenario is that the
RemoteAE3 to process ONCE each individual message.
   My questions are:

Q1. Is the configuration correct for my scenario? If NO, please provide
some configurations.

Q2. If the answers for the Q1 is YES, is the described behavior a known
problem of UIMA AS that has been solved in 2.4.0?


I look forward for your answers.
Thank you.
  Regards,
  Florin

IDs for annotations

Georg Fette — 2013-05-05T18:19:44

Hello,
Is there a way to give annotations (all structures derived from 
uima.tcas.Annotation) an ID member ? I tried to add a feature ID to the 
type Annotation but it is feature final. I could add such a feature to 
all subclasses of Annotation although that would result in a lot of 
different ID features for all direct subclasses of Annotation therfore 
being an ugly solution. Has somebody a nice solution for that ?
Greetings
Georg

Textmarker - Qualification of Types

2013-05-03T11:52:42

Hi,

I'm running Textmarker on a CAS XMI file with a lot of annotations from different annotators and different type systems. There are some type names used more than once, but with different name spaces. All types are defined in the type system included with TYPESYSTEM. Prepending the namespace to a type name in a Textmarker script does not work. How to tell Textmarker which namespace to use?

Thanks,
Armin