Entwicklung eines Konverterframeworks für linguistisch annotierte Daten auf Basis eines gemeinsamen (Meta-)modells

Florian Zipser (1, 2)

1 IDSL - Institut für Deutsche Sprache und Linguistik
2 Inria Saclay - Ile de France

In the field of linguistic research recent years have seen an increasing usage of linguistically annotated textual data referred to as corpora. As corpora become increasingly large the need for computational support increases as well. A variety of tools and formats for the creation of corpora have therefore been developed to investigate and store information about a wide range of linguistic phenomena. Unfortunately, most tools can only process their own proprietary formats and are often unable to import and export data in other desirable formats. To handle the consequent jungle of formats, I develop a framework to convert data coming from several formats into other formats. The framework contains the universal converter Pepper based on a common meta-model for linguistic annotated data called Salt. I present the two framework components and describe how the mappings from and to each of the investigated formats is achieved through the Salt meta-model.

Mots clés: Salt; Pepper; linguistic framework; universal converter; converter framework; linguistic data; linguistic annotation; syntactic annotation; morphologic annotation; PAULA; EXMARaLDA; ANNIS; Treetagger; TigerXML; meta-model; modeling; EMF; Eclipse Modeling Framework; OSGi; linguistic model; common model

Mots clés: linguistisches Datenmodell; linguistische Daten; Metamodell; gemeinsames Modell; Meta-Modell; Syntaxannotation; morphologische Annotation; Konverterframework

Domaines: Architectures Matérielles [cs.AR]; Informatique et langage [cs.CL]; Modélisation et simulation; Linguistique

Florian Zipser. Entwicklung eines Konverterframeworks für linguistisch annotierte Daten auf Basis eines gemeinsamen (Meta-)modells. 2009. ⟨hal-00606102⟩