Mapping Representation based on Meta-data and SPIN for Localization Workflows

Alan Meehan, Rob Brennan, Dave Lewis, Declan O'Sullivan
2014 Extended Semantic Web Conference  
The localization industry currently deploys language translation workflows based on heterogeneous tool-chains. Standardized tool interchange formats such as XLIFF (XML Localization Interchange File Format) have had some impact on enabling more agile translation workflows. However the rise of new tools based on machine translation technology and the growing demand for enterprise linked data applications has created new interoperability challenges as workflows need to encompass a broader range of
more » ... tools. In this paper we present an approach of representing mappings between RDF-based representations of multilingual content and meta-data. To represent the mappings, we use a combination of SPARQL Inferencing Notation (SPIN) and meta-data. Our approach allows the mapping representation to be published as Linked Data. In contrast to other frameworks such as R2R, the mappings are executed via a standard SPARQL processor. The objective is to provide a more agile approach to translation workflows and greater interoperability between software tools by leveraging the ongoing innovation in the Multilingual Web field. Our use case is a Language Technology retraining workflow where publishing mappings leads to new opportunities for interoperability and end-to-end tool-chain analytics. We present the results from an initial experiment which compared our approach of executing and representing mappings to that of a similar approachthe R2R Framework.
dblp:conf/esws/MeehanBLO14 fatcat:m4bblve6evbn5nydqhoo2ywjdm