Upgrade Illinois components

[reckart]

The Illinois Cognitive Computation Group has a comprehensive Maven repository containing their software and models: http://cogcomp.cs.uiuc.edu/m2repo/edu/illinois/cs/cogcomp/

Would be nice to upgrade the LBJ module in DKPro Core with the latest versions:

Organizational issues

• [x] Upgrade to CogComp 4.0.7
• [x] clarify licenses of the packages (cf. mailing list post)
• [x] Exclude transitive GPL dependency: Stanford CoreNLP which according to this post is used only for evaluation.
• [x] Exclude transitive GPL dependency: MySQL -- tests run without it
• [x] Exclude transitive GPL dependency: Weka -- tests run without it

Integrate components

• [X] Illinois Tokenizer / Sentence splitter -> IllinoisSegmenter, IllinoisStatefulSegmenter
• [x] Illinois Part of Speech Tagger (Model not yet configurable) -> IllinoisPosTagger
• [x] Illinois Named Entity Tagger -> IllinoisNamedEntityRecognizer
• [x] Illinois Lemmatizer -> IllinoisLemmatizer
• [x] Illinois Chunker -> IllinoisChunker

[mssammon]

Please incorporate only projects that are officially released via our software page (http://cogcomp.cs.illinois.edu/page/software/) or our github repo (https://github.com/IllinoisCogComp/illinois-cogcomp-nlp). The maven repository is a convenience that includes various works in progress that are not intended for general distribution. Thanks, Mark

[reckart]

@mssammon happy to comply with your wish. We will not make this module a part of an official release anyway unless the libraries would be available on Maven Central.

Right now, I have the following dependencies declared:

    <properties>
        <illinois-cogcomp-nlp.version>3.0.36</illinois-cogcomp-nlp.version>
        <illinois-cogcomp-srl.version>5.1.11</illinois-cogcomp-srl.version>
    </properties>
        <dependency>
          <groupId>edu.illinois.cs.cogcomp</groupId>
          <artifactId>illinois-pos</artifactId>
          <version>${illinois-cogcomp-nlp.version}</version>
        </dependency>
        <dependency>
            <groupId>edu.illinois.cs.cogcomp</groupId>
            <artifactId>illinois-chunker</artifactId>
            <version>${illinois-cogcomp-nlp.version}</version>
        </dependency>
        <dependency>
            <groupId>edu.illinois.cs.cogcomp</groupId>
            <artifactId>illinois-ner</artifactId>
            <version>${illinois-cogcomp-nlp.version}</version>
        </dependency>
        <dependency>
            <groupId>edu.illinois.cs.cogcomp</groupId>
            <artifactId>illinois-tokenizer</artifactId>
            <version>${illinois-cogcomp-nlp.version}</version>
        </dependency>
        <dependency>
            <groupId>edu.illinois.cs.cogcomp</groupId>
            <artifactId>illinois-lemmatizer</artifactId>
            <version>${illinois-cogcomp-nlp.version}</version>
        </dependency>
        <dependency>
            <groupId>edu.illinois.cs.cogcomp</groupId>
            <artifactId>illinois-srl</artifactId>
            <version>${illinois-cogcomp-srl.version}</version>
            <exclusions>
              <exclusion>
                <artifactId>ChuLiuEdmonds</artifactId>
                <groupId>edu.cmu.cs.ark</groupId>
              </exclusion>
            </exclusions>
        </dependency>

The software page (http://cogcomp.cs.illinois.edu/page/software_view/SRL) mentions SRL and also mentions a more recent version to be included with the curator. The SRL code appears not to be on GitHub.

Can it be integrated - in the version listed above or in another? Since the version above seems to depend on a SNAPSHOT version, it looks more like one of the convenience artifacts.

[reckart]

Please incorporate only projects that are officially released via our software page (http://cogcomp.cs.illinois.edu/page/software/) or our github repo (https://github.com/IllinoisCogComp/illinois-cogcomp-nlp). The maven repository is a convenience that includes various works in progress that are not intended for general distribution. Thanks, Mark

@mssammon I suppose this comment is no longer relevant since you advertise your Maven repo (under a new URL) via the GitHub page now?