UIMA-Based JCoRe 2.0 Goes GitHub and Maven Central ― State-of-the-Art Software Resource Engineering and Distribution of NLP Pipelines

UIMA-based JCORE 2.0 Goes GITHUB and MAVEN CENTRAL — State-of-the-Art Software Resource Engineering and Distribution of NLP Pipelines Udo Hahn1 Franz Matthies1 Erik Faessler1 Johannes Hellrich1 2 1 Jena University Language & Information Engineering (JULIE) Lab 2 Research Training Group “The Romantic Model. Variation - Scope - Relevance” Friedrich-Schiller-Universit¨at Jena Jena, Germany http://www.julielab.de Abstract We introduce JCORE 2.0, the relaunch of a UIMA-based open software repository for full-scale natural language processing originating from the Jena University Language & Information Engineering (JULIE) Lab. In an attempt to put the new release of JCORE on firm software engineering ground, we uploaded it to GITHUB, a social coding platform, with an underlying source code versioning system and various means to support collaboration for software development and code modification management. In order to automate the builds of complex NLP pipelines and properly represent and track dependencies of the underlying JAVA code, we incorporated MAVEN as part of our software configuration management efforts. In the meantime, we have deployed our artifacts on MAVEN CENTRAL, as well. JCORE 2.0 offers a broad range of text analytics functionality (mostly) for English-language scientific abstracts and full-text articles, especially from the life sciences domain. Keywords: natural language processing pipelines, open software repository, software engineering, software configuration man- agement, collaborative code management, JCORE, UIMA, GITHUB, MAVEN CENTRAL 1. Introduction A true sign of the growing maturity of natural language processing is the continuous production of reusable code (e.g., for POS tagging, parsing, WSD, reference resolution, etc.) and language data resources (e.g., annotated corpora or computational lexicons). For a long time, a common way to share such artifacts has been to create a publically accessible space on one’s own site, list available resources, and, after acknowledgment of licencing conditions, allow download of either source code or binaries. We have been following this passive directory model for many years with our own NLP tool suite, the JULIE Lab UIMA Component Repository (JCORE; Hahn et al. (2008)), as well. JCORE – based on the UIMA middleware framework1 – assembled (following UIMA speak) several collection readers (for document input), analysis engines (for NLP core tasks, e.g., document and sentence segmen- tation, named entity tagging or acronym resolution) and CAS consumers (e.g., for storing analysis results in an in- dex structure). These resources were hosted and made pub- lically accessible via http://www.julielab.de/ Resources/JCoRe+NLP+Tools.html. This service for the NLP community, despite the positive feedback we got from our users, increasingly also generated several problems which can be summarized as follows: • Some modules maintained on an irregular basis be- came outdated (e.g., the MEDLINE2 reader needs 1https://uima.apache.org/ 2MEDLINE is a bibliographical database for life sciences doc- uments which currently comprises more than 25M references; for more details, cf. https://www.nlm.nih.gov/bsd/ pmresources.html adaptation to the yearly updated MEDLINE XML schema), • Bug fixes from external users or functional extensions they asked for were hard to integrate in the daily work- flow. This is mainly due to the change patterns of the personnel working in academic software labs (e.g., …

Content truncated to 500 words