This is a project for building and running UIMA pipelines with JCoRe components. The jcore-pipeline-builder is a command line program that allows to easily choose JCoRe components to form a NLP pipeline. The components can be configured (input and output directories, algorithmical parameters, configuration file paths etc.) and the pipeline can be stored as a directory of files. The Java libraries required to run the pipeline are automatically downloaded as Maven artifacts and stored with the pipeline's descriptors and configuration files. This allows the jcore-pipeline-runner to run the created pipeline as a CPE or with UIMA DUCC. The pipeline runner takes in a configuration file in XML format. An empty version of such a configuration is created by the pipeline runner itself in case a non-existing file is passed to the program. The configuration file specifies how much memory and threads the pipeline should use. Alternatively, a UIMA DUCC deployment can be done. For build and usage information, please refer to the README in the GitHub repository.

While this project is primarily meant to be used with JCoRe components, third-party components could be included as well by delivering a so-called "meta description" of the component. This is a JSON file providing information required by the pipeline builder. Also, the component's code must be available via Maven, possibly local.