Báo cáo khoa học: "Computing Lattice BLEU Oracle Scores for Machine Translation" potx

... Association for Computational Linguistics, pages 120–129,Avignon, France, April 23 - 27 2012.c2012 Association for Computational LinguisticsComputing Lattice BLEU Oracle Scores for Machine TranslationArtem ... 4 and we use BLEU as a synonym for 4 -BLEU. BLEU is deﬁned for a pair of corpora, but, as an oracle decoder is working at the sentence-level, itshould rely on an approximation of BLEU that canlinear ... and RLX oracles werethe best performing oracles, with the ILP andRLX oracles being considerably faster, sufferingonly a negligible decrease in BLEU, compared tothe 4 -BLEU- optimized LB oracle. ...

Tài liệu Báo cáo khoa học: "Fast Decoding and Optimal Decoding for Machine Translation" doc

... Computer ScienceUniversity of Southern California Stanford University4676 Admiralty Way, Suite 1001 Stanford, CA 94305Marina del Rey, CA 90292 jahr@cs.stanford.edugermann,knight,marcu,kyamada @isi.eduAbstractA ... Fast Decoding and Optimal Decoding for Machine TranslationUlrich Germann , Michael Jahr , Kevin Knight , Daniel Marcu , and Kenji YamadaInformation Sciences Institute Department of ... of any statistical machine translation system. The decoder’s job isto ﬁnd the translation that is most likelyaccording to set of previously learnedparameters (and a formula for combin-ing them)....

Báo cáo khoa học: "Discriminative Modeling of Extraction Sets for Machine Translation" pptx

... Todd Ward, and Wei-Jing Zhu. 2002. BLEU: A method for automaticevaluation of machine translation. In Proceedings ofthe Annual Conference of the Association for Com-putational Linguistics.Andreas ... Association for Computational LinguisticsDiscriminative Modeling of Extraction Sets for Machine TranslationJohn DeNero and Dan KleinComputer Science DivisionUniversity of California, Berkeley{denero,klein}@cs.berkeley.eduAbstractWe ... Evan Herbst. 2007. Moses: Opensource toolkit for statistical machine translation. InProceedings of the Annual Conference of the Associ-ation for Computational Linguistics: Demonstrationtrack.Simon...

Báo cáo khoa học: "Learning Non-Isomorphic Tree Mappings for Machine Translation" pptx

... statistical formalisms (limited to isomorphictrees), synchronous TSG allows local distortion of the tree topol-ogy. We reformulate it to permit dependency trees, and sketchEM/Viterbi algorithms for ... naturally extend to train on pairs offorests (including packed forests obtained by chart pars-ing). The correct tree is presumed to be an element ofthe forest. This makes it possible to train ... estimate a model from unaligned data.4 A Probabilistic TSG Formalism For expository reasons (and to ﬁll a gap in the literature),ﬁrst we formally present non-synchronous TSG. Let Q bea set of...

Báo cáo khoa học: "FUNCTIONAL UNIFICATION GRAMMAR: A FORMALISM FOR MACHINE TRANSLATION" doc

... UNIFICATION GRAMMAR: A FORMALISM FOR MACHINE TRANSLATION Martin Kay Xerox Palo Alto Research Center 3333 Coyote Hill Road Palo Alto California 94304 and CSLI, Stanford Abstract Functional ... parts: a formal description of the relevant facts about the language, and an interpreter of the formalism. The formal description is data whereas the interpreter is program. The formal description ... engineering advantages are clear to see. A single formalism would straightfor- wardly reduce the number of interpreters to two, one for analysis and one for synthesis. Furthermore, the explanatory...

Báo cáo khoa học: "Disambiguating Temporal–Contrastive Discourse Connectives for Machine Translation" docx

... 46–51,Portland, OR, USA 19-24 June 2011.c2011 Association for Computational LinguisticsDisambiguating Temporal–Contrastive Discourse Connectives for Machine TranslationThomas MeyerIdiap Research Institute ... statistical machine translation sys-tems.1 IntroductionThe probabilistic phrase-based models used in sta-tistical machine translation (SMT) have been im-proved by integrating linguistic information ... 2008). For this resource, onehundred types of explicit connectives were manuallyannotated, as well as implicit relations not signaledby a connective. For French, the ANNODIS project for anno-tation...

Báo cáo khoa học: "Minimum Error Rate Training in Statistical Machine Translation" potx

... often used criteria are, for example,F-Measure for parsing, mean average precision for ranked retrieval, and BLEU or multi-reference worderror rate for statistical machine translation. The ... NIST and BLEU are accuracy measures,and thus larger values reﬂect better translation qual-ity. Note that NIST and BLEU scores are not addi-tive for different sentences, i.e. the score for a doc-ument ... is straightforward to reﬁne this algorithm toalso handle the BLEU and NIST scores instead ofsentence-level error counts by accumulating the rel-evant statistics for computing these scores (n-gramprecision,...

Báo cáo khoa học: "THE SPECIFICATION OF TIME MEANING FOR MACHINE TRANSLATION" pot

... interlingual representation, namely the combination after - before. (13) has the value contain - before, and (Ib) the value before - before. In (15) the situation is different: here, the time ... to yield a formula, e.g., W(yesterday(1))=1 iff the interval I is a subset of yesterday. 35 (iii) Temporal relations that take two intervals to yield a formula, e.g., W(beforeil,J))=1 ... for the assignment of deictic values in particular cases is the following one : in fact, an iterative interpretation, and for such interpretations we need a more complex representation format....

Báo cáo khoa học: "Improved Word-Level System Combination for Machine Translation" doc

... Similarly, full test set scores are ob-tained by accumulating counts over all hypothesisand reference pairs. The BLEU scores are betweenand , higher being better. Often BLEU scores arereported ... 41.88 51.45 68.62 BLEU tuned 42.12 51.72 68.59MTR tuned 54.08 38.93 71.42Table 1: Mixed-case TER and BLEU, andlower-case METEOR scores on Arabic NISTMT03+MT04.Arabic test TER BLEU MTRsystem ... opti-mize BLEU. Again, the best scores on each met-ric are obtained by the combination tuned for thatmetric. Only the METEOR score of the TER tunedcombination is worse than the METEOR scores...

Báo cáo khoa học: "Pattern-Based Context-Free Grammars for Machine Translation" pot

... of the World-Wide Web (WWW) as information source, it has become rou- tine for Internet users to access textual data written in foreign languages. In Japan, for example, a dozen or so inexpensive ... systems based on these formalisms, on the other hand, would not be possible without much more efficient parsing and disambiguation algorithms for these formalisms and a method for building a lexicon ... Unification-based grammar for- malisms and lexical-semantics formalisms (see LFG (Kaplan and Bresnan, 1982), HPSG (Pollard and Sag, 1987), and Generative Lexicon (Pustejovsky, 1991), for example) have...

Xem thêm