Báo cáo khoa học: "Lattice Parsing to Integrate Speech Recognition and Rule-Based Machine Translation" pdf

... April 2009.c2009 Association for Computational LinguisticsLattice Parsing to Integrate Speech Recognition and Rule-Based Machine TranslationSelçuk KöprüAppTek, Inc.METU TechnopolisAnkara, ... novel approach to integrate speech recognition and rule-based machine translation by lattice pars-ing. The presented approach is hybridin two senses. First, it combines struc-tural and statistical ... by the parser can be designed top-down,bottom-up or in a combined manner. It employsan agenda to store the edges prior to inserting to the chart. Edges are deﬁned to be either completeor incomplete....

Báo cáo khoa học: "Discriminative Strategies to Integrate Multiword Expression Recognition and Parsing" docx

... beeasily integrated in the MWE sequence. Constant and Sigogne (2011) proposed to combine MWE seg-mentation and part-of -speech tagging into a singlesequence labelling task by assigning to each token ... parametervector θ and a feature vector f:Vθ(p) = θ.f(p) =mj=1θj.fj(p)where fj(p) corresponds to the number of occur-rences of the feature fjin the parse p. According to 206Charniak and ... inFrench. We also integrated mixed bigrams made upof a word and a part-of -speech. Speciﬁc features. Due to their different use, eachmodel integrates some speciﬁc features. In order to deal with...

Báo cáo khoa học: "Dependency Parsing of Hungarian: Baseline Results and Challenges" potx

... 933–939.Szilárd Iván, Róbert Ormándi, and András Kocsor.2007. Magyar mondatok SVM alapú szintaxiselemzése [SVM-based syntactic parsing of Hun-garian sentences]. In V. Magyar ... Conference on Language Resources and Evaluation (LREC ’06).Dániel Varga, Péter Halácsy, András Kornai, ViktorNagy, László Németh, and Viktor Trón. 2005. Par-allel corpora ... one to- ken in the Szeged Dependency Treebank. Assum-ing a perfect phrase recogniser on the whitespacetokenised input for them is quite unrealistic. Thuswe decided to split them into tokens...

Báo cáo khoa học: "Attacking Parsing Bottlenecks with Unlabeled Data and Relevant Factorizations" pdf

... sibling and grandparent factorizationsdescribed above–for Conversion 1, sibling scoringmay help conjunctions and grandparent scoring mayhelp prepositions, and for Conversion 2, grandparentscoring ... (forprepositions) will lead to the largest improvements.For the phrase dogs and cats, edge-based countswould measure the associations between dogs and and, and and and cats, but never any web ... factorization (Grand+Sib) and the version which uses just the grandparent fac-torization (Grand). A parser which uses only grand-parents (referred to as Model 0 in Koo and Collins(2010)) may...

Báo cáo khoa học: From meiosis to postmeiotic events: Alignment and recognition of homologous chromosomes in meiosis ppt

... transcribed rDNA loci and histonegenes [36]. A previous model proposed roles fortranscription and for a specialized transcription fac-tory in homologous chromosome recognition and pairing [37,38]. ... attached to a speciﬁctranscription factory, in which transcriptionalmachinery proteins are aggregated, and those DNAregions that are not undergoing transcription pro-trude from the factory and ... chromosome-speciﬁc recognition proteins are linked to cytoskeletalmotor proteins to tether homologous chromosomes.In Schiz. pombe, we recently uncovered a novel phe-nomenon relating noncoding RNA to homologouschromosome...

Tài liệu Báo cáo khoa học: "Modiﬁed Distortion Matrices for Phrase-Based Statistical Machine Translation" doc

... the decoder to modify the distortionmatrix just before starting the search. As usual, thedistortion matrix is queried by the distortion penaltygenerator and by the hypothesis expander9.7.1 ... chunk-based rules following Bisazza and Federico (2010).Shallow syntax chunking is indeed a lighter and simpler task compared to full parsing, and it canbe used to constrain the number of reorderings ... sen-tence varies according to the rule set, to the size ofchunks and to the context. A high degree of fuzzi-ness can complicate the decoding process, leavingtoo much work to the in-decoding reordering...

Tài liệu Báo cáo khoa học: "Improving Word Representations via Global Context and Multiple Word Prototypes" pdf

... with vectors that capture semantic and syntac-tic information of words. These representations canbe used to induce similarity measures by computingdistances between the vectors, leading to many ... each word is only repre-sented with one vector, which clearly fails to capturehomonymy and polysemy. Reisinger and Mooney(2010b) introduced a multi-prototype VSM whereword sense discrimination ... clus-tering contexts, and then prototypes are built usingthe contexts of the sense-labeled words. However, inorder to cluster accurately, it is important to captureboth the syntax and semantics of...

Tài liệu Báo cáo khoa học: "Unsupervised Search for The Optimal Segmentation for Statistical Machine Translation" doc

... correspond to different random orders inprocessing the data (Creutz and Lagus, 2007).are affected by the decision in the current step.This leads to a sequential search and does not lenditself to ... as in Creutz and Lagus (2007). To sum-marize brieﬂy, the prior P (Mf) is assumed to onlydepend on the frequencies and lengths of the indi-vidual morphs, which are also assumed to be in-dependent. ... translation taskat hand. Existing solutions to this problem arepredominantly heuristic, language-dependent, and as such are not easily portable to other lan-guages. Another point to consider is...

Tài liệu Báo cáo khoa học: "Topological Ordering of Function Words in Hierarchical Phrase-based Translation" pdf

... Proceedings of the 47th Annual Meeting of the ACL and the 4th IJCNLP of the AFNLP, pages 324–332,Suntec, Singapore, 2-7 August 2009.c2009 ACL and AFNLP1 2 3 1123{ }324X → γ, α, ∼X...

Tài liệu Báo cáo khoa học: "Arabic Tokenization, Part-of-Speech Tagging and Morphological Disambiguation in One Fell Swoop" pdf

... getstokenized correctly, independently of the number ofresulting tokens; the token-based measures refer to the four token ﬁelds into which the ATB splits eachworddetermines the ATB tokenization. ... corpora asTR1 and TR2, and to the test corpora as, TE1 and TE2. We report results on both TE1 and TE2 be-cause of the differences in the two parts of the ATB,both in terms of origin and in terms ... request.575velopment, training, and test corpora with roughly12,000 word tokens in each of the development and test corpora, and 120,000 words in each of the train-ing corpora. We will refer to the training...

Xem thêm