Báo cáo khoa học: "Cross-Framework Evaluation for Statistical Parsing" pptx

... ap- propriate for a formalism like CCG with a large set of lexical categories than one generic token for all unknown words. The performance of the baseline model is shown in the top row of table 3. For ... order to compare our performance with the parser of Clark et al. (2002), we also evaluate our best model according to the dependency evaluation introduced for that parser. For f...

Ngày tải lên: 31/03/2014, 06:20

... the Association for Computational Linguistics:shortpapers, pages 294–298, Portland, Oregon, June 19-24, 2011. c 2011 Association for Computational Linguistics Corpus Expansion for Statistical Machine ... SSRs. Notice that for each new sentence generated, we al- low for application of only one substitution. Although the idea is straightforward, we face two problems in practice. Fi...

Ngày tải lên: 20/02/2014, 04:20

... smorgasbord of features for statistical machine translation. In Proceedings of HLT- NAACL 2004. Och, F. J., Tillmann, C., and Ney, H. (1999). Improved align- ment models for statistical machine translation. ... 2005. c 2005 Association for Computational Linguistics Clause Restructuring for Statistical Machine Translation Michael Collins MIT CSAIL mcollins@csail.mit.edu Philipp...

Ngày tải lên: 20/02/2014, 15:20

... for the experiments reported here is the same as the one used for other experiments reported in this paper. The results in Table 3 illustrate how the language model performs reasonably well for ... search- ing for the optimal solution among all possible permu- tations computationally intractable. Therefore, SMT decoders typically limit the number of permutations considered for effici...

Ngày tải lên: 08/03/2014, 02:21

... Chinese–English translation as performed in the NIST MT evaluation (www.nist.gov/speech/tests/mt). For the French–English translation task, we used the EuroParl corpus as distributed for the shared task in ... the decoder, and the evaluation is done on the test set provided for the NAACL 2006 shared task. For the Chinese–English translation task, we used the corpora distributed f...

Ngày tải lên: 08/03/2014, 02:21

... in- crease in performance of about 2-3% for most mod- els. For General-LM, results increased by 8-10%. The Poisson and general fertility models show a 2- 5% gain in performance over the basic ... English. For the ATIS task, our formal language is a mi- nor variant of the NL-Parse (Hemphill, Godfrey, and Doddington, 1990) used by ARPA to annotate the ATIS corpus. An example of a f...

Ngày tải lên: 08/03/2014, 21:20

... dev2 and dev3) for development and testing. We used the dev1 set for parameter tuning, the dev2 set for choosing the setting of the proposed method, which is described below, and the dev3 set for test- ing. The ... recognition. Thus, the translation quality for lattice inputs is better than the quality for 1- best inputs. In this paper, we show that lattice decoding is also useful...

Ngày tải lên: 17/03/2014, 00:20

... avoid the unwanted bias toward small clusters. For cluster , the weight for sample is proportional to . Weight by Performance: The idea of weight by performance is to focus the model on its weakness when ... that for about the same pars- ing accuracy, we only need to annotate a third of the samples as compared to the usual random selection method. 1 Introduction A prerequisite for bui...

Ngày tải lên: 17/03/2014, 08:20

... Meeting of the Association for Computational Linguistics, pages 604–611, Uppsala, Sweden, 11-16 July 2010. c 2010 Association for Computational Linguistics Error Detection for Statistical Machine Translation ... incorrect parts from correct parts is therefore very desir- able not only for post-editing and interactive ma- chine translation (Ueffing and Ney, 2007) but also for SMT i...

Ngày tải lên: 23/03/2014, 16:20

... to minimize some information loss such as the KL divergence KL(p  q). The simpler model q can then act as a surrogate for p during inference. 3.2 Variational Decoding for MT For each input sentence ... q ∗ (· | h) constrained to be a normalized distribution for each h. 596 Brute-Force-MLE(HG(x )) 1 for y, d in HG(x) ✄ each derivation 2 for w in y ✄ each n-gram type 3 ✄ accumulate...

Ngày tải lên: 23/03/2014, 16:21