Erferenheter av Swetra— ett svenskt MT-experiment

Bengt Sigurd

Research output: Chapter in Book/Report/Conference proceedingPaper in conference proceeding

Abstract

Swetra is primarily a research project, but its computer programs and
reports may result in commercial products. The research group at Lund
consists of a few persons only, which has advantages eind disadvantages.
The main languages studied are Russian, English and Swedish. The grammatical model is Referent Grammar (RG) and the project is a spin-off
of the development of that grammar. Referent Grammar is inspired by
generalized phrase structure grammar (GPSG) and written directly in
Prolog (DCG), which is why the grammars can run directly on computers.
Arity Prolog and PC’s are used. The work within Swetra consists mainly
of writing grammar modules and lexicons for different languages. These
modules are connected in automatic translation. Referent grammars are
bidirectional and can be used both for analysis and synthesis (generation).
A certain number of transfer rules are also needed in translation, however,
in order to make necessary changes of functional representations, add
features such as definiteness when translating from Russian into English
etc. The paper gives a survey of the experience from Swetra reseeirch so
far.
Original languageSwedish
Title of host publicationPapers from the Seventh Scandinavian Conference of Computational Linguistics
Subtitle of host publicationReykjavík 1989
EditorsJörgen Pind, Eiríkur Rögnvaldsson
Place of PublicationReykjavík
PublisherUniversity of Iceland
Pages229-236
Number of pages8
Publication statusPublished - 1990
EventProceedings of the 7th Nordic Conference of Computational Linguistics - University of Iceland, Reykjavík, Iceland
Duration: 1989 Jun 261989 Jun 28
Conference number: 7
https://aclanthology.org/volumes/W89-01/

Conference

ConferenceProceedings of the 7th Nordic Conference of Computational Linguistics
Abbreviated titleNODALIDA 1989
Country/TerritoryIceland
CityReykjavík
Period1989/06/261989/06/28
Internet address

Subject classification (UKÄ)

  • General Language Studies and Linguistics

Cite this