MOLTO poster for META Forum, Brussels 2010, Belgium.

272 views
209 views

Published on

MOLTO is funded by the European Union Seventh Framework Programme (FP7/2007-2013) under grant agreement FP7-ICT-247914. MOLTO is a project about high-quality multilingual online translation. http://molto-project.eu

Published in: Education
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
272
On SlideShare
0
From Embeds
0
Number of Embeds
1
Actions
Shares
0
Downloads
0
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

MOLTO poster for META Forum, Brussels 2010, Belgium.

  1. 1. MOLTO’s goal is to develop tools for web content providers to translate texts between multiple languages in realtime with high quality. Languages are separate modules in the tool and can be varied; prototypes covering amajority of the EU’s 23 official languages will be built.GF - Grammatical FrameworkThe core of a MOLTO translation system is a multilingual GF grammarwhere meaning-preserving translations are obtained as composition ofparsing and generation via the abstract syntax, an interlingua. GF is aframework for interlinguas in which the basic linguistic details oflanguages, inflectional morphology and syntactic combinationfunctions, are provided via the Resource Grammar Library.MOLTO will further improve grammar engineering in GF by:❖ Integrated Development Environment to use the RGL and tomanage large projects;❖ Example-based grammar writing support to bootstrap a grammarfrom a set of example translations.Statistical Machine TranslationMOLTO will develop and evaluate combination approaches tointegrate grammar-based and SMT models in a hybrid, robust MTsystem. At least four variants will be studied:❖ baseline: cascade of independent MT systems;❖ hard integration: GF partial output is fixed in a regular SMTdecoding;❖ soft integration I: GF partial output, as phrase pairs, is integratedas a discriminative probability feature model in a phrase-basedSMT system;❖ soft integration II: GF partial output, as tree fragment pairs, isintegrated as a discriminative probability model in a syntax-basedSMT system.OWL OntologiesMOLTO sees ontologies as a way to formalize interlinguas in specificdomains. Based on this observation, it will carry out research todevelop two-way grammar-ontology interoperability that will bridgenatural language and formal knowledge. The resulting MOLTOinfrastructure will allow knowledge modeling, semantic indexing andretrieval using natural language. The engine will perform semi-automatic creation of abstract grammars from ontologies; deriveontologies from grammars, and retrieve instance level knowledge from/in natural language by first transforming queries to semantic queries,and secondly by expressing the resulting knowledge in naturallanguage.molto-project.eutwitter.com/moltoprojectM LTOOnon multa, sed multumMultilingual On-line TranslationTools and Resourcesdelivered by MOLTOwill be distributed alsothroughFP7- 247914 at a glanceICT-2009.2.2 - Language based interactionDuration: 36 monthsStart date: 1 March 2010End date: 28 February 2013EU funding: 2.3 Millions €HowFar : Place -> Question;Zoo : PlaceKind;HowFar(Zoo)Zoo = mkPlace (mkN "zoo" masculine) dative;HowFar place = mkQS(mkQCl what_distance_IAdv place.name);À quelle distance est le zoo?Zoo = mkPlace (mkN "djurpark" "djurparker") "i";HowFar place = mkQS(mkQCl far_IAdv(mkCl(mkVP place.to)));Hur långt är det till djurparken?parse linearize

×