Building an Owl-Ontology for Representing, Linking and Querying SemAF Discourse Annotations

Abstract

Linguistic Linked Open Data (LLOD) are technologies that provide a powerful instrument for representing and interpreting language phenomena on a web-scale. The main objective of this paper is to demonstrate how LLOD technologies can be applied to represent and annotate a corpus composed of multiword discourse markers, and what the effects of this are. In particular, it is our aim to apply semantic web standards such as RDF and OWL for publishing and integrating data. We present a novel scheme for discourse annotation that combines ISO standards describing discourse relations and dialogue acts – ISO DR-Core (ISO 24617-8) and ISO-Dialogue Acts (ISO 24617-2) in 9 languages (cf. Silvano and Damova 2022; Silvano, et al. 2022). We develop an OWL ontology to formalize that scheme, provide a newly annotated dataset and link its RDF edition with the ontology. Consequently, we describe the conjoint querying of the ontology and the annotations by means of SPARQL, the standard query language for the web of data. The ultimate result is that we are able to perform queries over multiple, interlinked datasets with complex internal structure. This is a first, but essential step, in developing novel, powerful, and groundbreaking means for the corpus-based study of multilingual discourse, communication analysis, or attitudes discovery.

Keywords

• LLOD, OWL ontology
• RDF
• discourse markers
• ISO standard
• annotations
• parallel corpus