| 1 |
SEM-I rational MT. Enriching deep grammars with a semantic interface for scalable machine translation
|
|
|
|
In: http://www.kecl.ntt.co.jp/icl/mtg/members/bond/pubs/2005-summit-semi.pdf (2005)
|
|
BASE
|
|
Show details
|
|
|
| 2 |
SEMI rational MT — enriching deep grammars with a semantic interface for scalable machine translation
|
|
|
|
In: http://www.mt-archive.info/MTS-2005-Flickinger.pdf (2005)
|
|
BASE
|
|
Show details
|
|
|
| 3 |
The Interplay Between Lexical and Syntactic Resources in Incremental
|
|
|
|
In: http://www.lrec-conf.org/proceedings/lrec2014/pdf/1085_Paper.pdf
|
|
BASE
|
|
Show details
|
|
|
| 4 |
ParGramBank: The ParGram Parallel Treebank
|
|
|
|
In: http://aclweb.org/anthology/P/P13/P13-1054.pdf
|
|
BASE
|
|
Show details
|
|
|
| 5 |
1 Deriving Thesaurus Entries from Translational Features
|
|
|
|
Abstract:
Dyvik (1998a-b and 2002) presents a method for deriving lexical sense divisions, and feature descriptions of each sense, automatically from translational data taken from a parallel corpus (ENPC). The senses are grouped in semantic fields based on shared translational properties. The feature sets assigned to the senses in a field form a semilattice based on inclusion and overlap relations among the feature sets. An example of such a derived semilattice is shown below: According to the hypothesis behind the method the dominance relations in the lattice express hyperonymy/hyponymy relations among the senses, and distance in the lattice generally reflects semantic distance. Hence the lattices should contain some of the information generally found in word nets and thesauri. This is restricted to relations of semantic similarity (hyperonymy/hyponymy and synonymy), while, e.g., antonymy and meronymy probably cannot be derived from the translational data. The present paper discusses the further derivation of thesaurus entries from the information in the feature lattices. Much of the detail in the derived lattices is probably the result of accidental biases and gaps in the parallel corpus from which they were derived. The derivation of a thesaurus entry, in which words are grouped as 'hyperonyms', 'hyponyms', 'synonyms ' and 'related words ' of an entry, therefore needs to abstract away from some of the detail in the lattices, and this can obviously be done in more than one way. The following entry for the Norwegian noun 'rett ' has been derived automatically from the translationally based lattices by the principles to be described. Sense 4 corresponds to the lattice shown above: 2 Extended bases
|
|
Keyword:
dish.) Synonyms; fat; føde; kar; kosthold; måltid; matvare; middag. Related words; næring. Subsense (ii) (Translation; Synonyms
|
|
URL: http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.148.9887
|
|
BASE
|
|
Hide details
|
|
|
| 6 |
TRANSLATIONS AS A SEMANTIC KNOWLEDGE SOURCE Abstract
|
|
|
|
In: http://www.hf.uib.no/i/LiLi/SLF/und/DASP203/Dyvik.pdf
|
|
BASE
|
|
Show details
|
|
|
| 7 |
1 Translations as Semantic Mirrors: From Parallel Corpus to Wordnet 1
|
|
|
|
In: http://www.hf.uib.no/i/LiLi/SLF/ans/Dyvik/ICAMEpaper.pdf
|
|
BASE
|
|
Show details
|
|
|
| 8 |
Linguistically Motivated Parallel Parsebanks
|
|
|
|
In: http://tlt8.unicatt.it/FullPaper/Poster_4.pdf
|
|
BASE
|
|
Show details
|
|
|
|
|
