Annotation Interoperability for the Post-ISOCat Era

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 5668–5677 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 5668 Annotation Interoperability for the Post-ISOCat Era Christian Chiarcos, Christian F¨ath, Frank Abromeit Applied Computational Linguistics Lab (ACoLi) Goethe University Frankfurt, Germany {chiarcos, faeth, abromeit}@em.uni-frankfurt.de Abstract With this paper, we provide an overview over ISOCat successor solutions and annotation standardization efforts since 2010, and we describe the low-cost harmonization of post-ISOCat vocabularies by means of modular, linked ontologies: The CLARIN Concept Registry, LexInfo, Universal Parts of Speech, Universal Dependencies and UniMorph are linked with the Ontologies of Linguistic Annotation and through it with ISOCat, the GOLD ontology, the Typological Database Systems ontology and a large number of annotation schemes. Keywords: interoperability, ontologies, linked data, linguistic annotation 1. Background The divergence and heterogeneity of linguistic annotations even for comparable language resources for the same lan- guage variety has long been recognized as a key problem in the advancement of human language technology. The cre- ation of language resources and tools is a laboursome and cost-intense process, but if these cannot be easily combined with each other, be it to feed the output of one annotator (say, a POS tagger) as input to another (say, a parser), or to increase the amount of available training data by combining multiple corpora of the same kind (say, morphology, part of speech annotations, dependency syntax or phrase structure syntax), potential synergies between those annotation ef- forts cannot be exploited and progress in the field is thus unnecessarily delayed. Very often, this is the situation for low-resource languages, where no standard annotation for a particular phenomenon has been established yet, but dif- ferent researchers develop multiple schemas independently from each other. With ISOCat,1 the ISO TC37 Data Category Registry (DCR), a possible solution to this problem has been pro- posed in the 2000s, a central repository of linguistic ter- minology managed in the context of a community process with a relatively low entry barrier, where individual re- searchers or institutions would just register their resource- or application-specific terminology. ISOCat focused on the elementary level of linguistic terms, and deliberately excluded the relations between these terms, but eventu- ally, it was imagined that these relations could be added in a separate relation registry (Schuurman and Windhouwer, 2011), and the process to develop these relations and terms would converge towards a consistent terminological inven- tory whose content would be the basis for a subsequent standardization process within ISO TC 37. In the standardization approach, terminology harmoniza- tion is achieved by aggregation and consolidation across all possible user communities within a centralized, monolithic repository. An alternative approach emerging simultane- ously to ISOCat was the idea of distributed terminology harmonization by creating links among independent, self- contained domain terminologies and between them and one 1http://www.ISOCat.org/ or multiple ‘upper models’, especially by means of ontolo- gies and Semantic Web technologies:2 Different user com- munities formalize and provide their respective terminol- ogy in a stand-alone, …

Content truncated to 500 words