A Conceptual Framework of Online Natural Language Processing Pipeline Application
Chunqi Shi, James D. Pustejovsky, Marc Verhagen · 2014
This paper describes a conceptual framework that enables online NLP pipelined applications to solve various interoperability issues and data exchange problems between tools and platforms; e.g., tokenizers and part-of-speech taggers from GATE, UIMA, or other platforms.We propose a restful wrapping solution, which allows for universal resource identification for data management, a unified interface for data exchange, and a light-weight serialization for data visualization.In addition, we propose a semantic mapping-based pipeline composition, which allows experts to interactively exchange data between heterogeneous components.