DUUI: A Toolbox for the Construction of a new Kind of Natural Language Processing

Giuseppe Abrami, Daniel Baumartz, Alexander Mehler

Abstract: Today, the heterogeneity of NLP tools in relation to existing methods and the constantly growing availability of models confronts various disciplines with major challenges in the daily handling of natural language processing. These are golden times for all scientific fields, as different models can be applied to the respective corpora; although in the short term this leads to non-trivial challenges in terms of analysis time, heterogeneity of (corpora) formats, processing input and output formats as well as analyzeability. These many construction phases show the need for a reliable working tool that can be used without intensive training, which is available in the form of Docker Unified UIMA Interface (DUUI). DUUI is designed as a platform-independent annotation framework for the horizontal and vertical distribution of heterogeneous NLP processes towards microservice-oriented homogenization using web services for a unified processing and reuse of unstructured data.

External IDs:dblp:conf/dhd/AbramiBM25