Tài liệu Báo cáo khoa học: "Improving Statistical Machine Translation with Monolingual Collocation" pdf

... 11-16 July 2010. c 2010 Association for Computational Linguistics Improving Statistical Machine Translation with Monolingual Collocation Zhanyi Liu 1 , Haifeng Wang 2 , Hua Wu 2 , Sheng Li 1 ... paper proposes to use monolingual collocations to improve Statistical Ma- chine Translation (SMT). We make use of the collocation probabilities, which are estimated from monol...

Ngày tải lên: 20/02/2014, 04:20

Ngày tải lên: 21/02/2014, 20:20

... leads to a mistranslation in a machine translation system. Therefore, the issue becomes how to parse tele.graphic messages accurately and efficiently to produce high quahty translation output. ... Misparsing re- duced by omissions has a far-reaching consequence in machine translation. Namely, a misparse of the input often leads to a translation into the target language which...

Ngày tải lên: 22/02/2014, 03:20

... system within the computer-aided Saarbriicken Translation System (STS), i.e. in human-aided MT and in machine- aided human translation. Titles of scientific papers from German databases were machine- translated ... sublanguage notion for disambi- guation and the selection of target language equivalents in machine translation. In this paper a theoretical concept and its imple-...

Ngày tải lên: 22/02/2014, 10:20

... the transliteration pairs. We propose a second model p 2 (e, f ) to deal with non-transliteration pairs (the “non-transliteration model”). Interpolation with the non-transliteration model allows the transliteration model ... initialized with a uniform distribu- tion and λ is set to 0.5. The expected count of a multigram q (E-step) is computed by multiplying the posterior probability of...

Ngày tải lên: 19/02/2014, 19:20

... learning algorithms and as extra word features in NLP systems. However, most of these models are built with only local context and one represen- tation per word. This is problematic because words are ... accounts for homonymy and polysemy by learning mul- tiple embeddings per word. We introduce a new dataset with human judgments on pairs of words in sentential context, and evaluate our mo...

Ngày tải lên: 19/02/2014, 19:20

... information of sub-trees in a given parse. With help of these new features, our sys- tem achieves 93.49 F-measure with hand-crafted parses. Comparison with the best reported results, 92.0 (Xue, ... arguments of a predicate are la- beled with a contiguous sequence of integers, in the form of AN (N is a natural number); the ad- juncts are annotated as such with the label AM followed b...

Ngày tải lên: 20/02/2014, 04:20

... instances with that label as positive instances and instances with any other label as negative instanc- es. The final class label is assigned by choosing the class that was assigned with the ... its performance with our original system. 4.1 Data The training set includes 349 clinical notes, with 11,967 assertions of medical problems. The test set includes 477 texts with 18...

Ngày tải lên: 20/02/2014, 05:20

... Li, 2007). 1 Even with all of these, however, there remains a significant gap between this WER and the threshold of 25%, at which lec- ture transcripts have been shown with statistical significance ... you ⇓ Output all rules for replacing the incorrect ASR sequence with the correct text, using the entire sequence (a) or splices (b), with or without surrounding anchors: (a) the okay on...

Ngày tải lên: 20/02/2014, 07:20

... On the other hand, the system with preceding infor- mation is not significantly better than the system without it 5 . Other non-local information may im- prove performance with our framework and this ... of the classifier on development data is 74.64 (without preceding information) and 75.14 (with preceding information). 470 Table 5: Performance with filtering on the development data. (&l...

Ngày tải lên: 20/02/2014, 12:20