Skip to content

JULIELab/jcore-pipelines

Repository files navigation

JCoRe Pipelines

These are pre-built JCoRe pipelines for various NLP aspect as described in (please cite this paper if you publish results using JCoRe):

Udo Hahn, Franz Matthies, Erik Faessler and Johannes Hellrich: UIMA-Based JCoRe 2.0 Goes GitHub and Maven Central ― State-of-the-Art Software Resource Engineering and Distribution of NLP Pipelines. In: Nicoletta Calzolari (Conference Chair), Khalid Choukri, Thierry Declerck, Marko Grobelnik, Bente Maegaard, Joseph Mariani, Asuncion Moreno, Jan Odijk, Stelios Piperidis (Eds.): Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC 2016), 2016. Portorož, Slovenia. [Full Text]

Objective

Requirements & Dependencies

In order to use these pipelines you need at least JDK 7(Java SE Development Kit 7), UIMA 2.6 & Maven 3.0. We also recommend using Eclipse IDE for Java Developers as your Java IDE if you're planning on using/building the pipelines programmatically(?) with Java, since it comes with the Maven Plugin. However, you're free to try it with different versions than those mentioned, but we can't make promises for a flawless functioning of our components in these cases.

Java method

The two Pipelines jcore-medical-pos-pipeline & jcore-relation-extraction-pipeline use the runPipeline.sh script to download all necessary maven artifacts and compile the Java files for starting the pipeline. That is, you will only need to install Maven and Java.
They differ in the method the Java files utilize UIMA:

  • Whereas the first mentioned pipeline just creates and runs a Pipeline with UIMAfit (so you basically have only the option to modify the Pipeline like you would with a CPE approach) ...
  • the second implementation would you allow to further augment your pipeline and its intermediate results with everything Java & UIMA hast to offer.
    Please see the respective Java files in src/main/java for the examples. If you change any of the Java files, be sure to run mvn clean and after that ./runPipeline.sh again (if you didn't change any dependencies in the file and/or the pom.xml it should also be possible to just delete the target/classes folder and run mvn compile).

CPE method - Plug'n'Play

The jcore-named-entity-pipeline employs the cpe-pipeline.xml to configure and run a Pipeline. You will need everything installed and properly configured that is mentioned under Requirements & Dependencies.
To check if your system is correctly configured and download all dependencies run ./installComponents.sh if no errors are encountered you're good to go to ./runPipeline.sh after you either put your files into data/inFiles or configure the cpe-pipeline.xml file to use another input folder.