A stand-off XML-TEI representation of reference annotation
Résumé
In this poster, we present an XML-TEI conformant stand-off representation of reference in discourse, building on the seminal work
carried out in the MATE project (Poesio, Bruneseaux & Romary 1999) and the earlier proposal on a reference annotation framework in Salmon-
Alt & Romary (2005). We make a three-way distinction between markables (the referring expressions), discourse entities (referents in the textual or extra-textual world), and links (relations that hold between referents, e.g., part-whole). Our approach differs from previous suggestions in that (i) inherent properties of the referent itself (e.g., animacy) are disentangled from the expressions used to refer to that referent, (ii) existing annotations from other layers such as morphosyntax are cleanly separated from the annotation of reference, but can be combined in queries and (iii) our
proposal is integrated into the larger structure of existing TEI-ISO standards, thereby allowing for compatibility with existing TEI-encoded
corpora and data sustainability. The workflow of adding reference annotations to an existing corpus will
be demonstrated with concrete examples from ongoing work in the SFB 1252 (subprojects C01 and INF), where this representation of
reference is the backbone for the annotation of (sentence) topic chains in dialogue data and for queries of topics in various grammatical
constructions.
Domaines
Informatique et langage [cs.CL]
Fichier principal
DGfS18_Adli_Engel_Romary_Same_draft-04-03-18.pdf (2.25 Mo)
Télécharger le fichier
DGfS-Poster-18_paper_7.pdf (88.28 Ko)
Télécharger le fichier
Origine | Fichiers produits par l'(les) auteur(s) |
---|
Origine | Fichiers produits par l'(les) auteur(s) |
---|
Loading...