Linked Data-Based NLP Workflows

Abstract

In this chapter we describe principles and architectures that support the development of NLP workflows and pipelines based on linked data technology. The benefit of NLP workflows that build on linked data standards is that they build on an open set of data models and Web technologies that can be implemented with standard functionality not requiring additional frameworks and thus avoiding any type of lock-in or dependence on particular frameworks in comparison to using UIMA, GATE or other frameworks. In this chapter we describe, on the one hand, how NLP workflows can be implemented by relying on the Natural Language Processing Interchange Format (NIF). We give examples of how a POS-tagger and a dependency parser can be implemented as NIF-based web services. We then describe Teanga, a recent platform for NLP integration that exploits Docker containers to implement NLP workflows. Finally, we also describe LAPPS Grid, an open-source platform for NLP tools that builds on JSON-LD.