Báo cáo khoa học: "Word Alignment via Submodular Maximization over Matroids" pot

... Oregon, June 19-24, 2011. c 2011 Association for Computational Linguistics Word Alignment via Submodular Maximization over Matroids Hui Lin Dept. of Electrical Engineering University of Washington Seattle, ... word alignment problem as maximizing a submodular function subject to matroid constraints (to be defined in Section 2). Submodular objective functions can represent comp...

Ngày tải lên: 23/03/2014, 16:20

... word alignment that uses synonym information as a regularization term. The experimental results show that our proposed method significantly improves word alignment quality. 1 Introduction Word alignment ... word 137 Figure 1: Graphical model of HM-BiTAM alignment quality. 2 Bilingual Word Alignment Model In this section, we review a conventional gener- ative word alignment model, H...

Ngày tải lên: 20/02/2014, 04:20

... et al. (2005) took all alignment links as sure links. If we use to represent the set of alignment links identified by the proposed methods and to denote the reference alignment set, the meth- ods ... additional corpora and with L3 as the pivot language, we build a word alignment model for L1 and L2. This approach can build a word alignment model for two languages even if no...

Ngày tải lên: 20/02/2014, 12:20

... combine word alignments over multiple mono- lingually motivated word segmentation. Our approach is based on link confidence score de- fined over multiple segmentations, thus the combined alignment ... segmentation optimi- zation based on alignment or postponing segmenta- tion combination late till SMT decoding phase, we try to combine word alignments over multiple monolingually...

Ngày tải lên: 23/03/2014, 16:20

... three filters Multiple Alignment Selection filter, Best Alignment Score Selection filter and Frequency Range constraint to the raw results to increase the accuracy of alignment. The Multiple Alignment Selection(MAS) ... from the alignment point of view. Somers has further observed that if the Best Alignment Score Selection (BASS) filter is ap- plied to yield the first few best results...

Ngày tải lên: 23/03/2014, 18:20

... further gains again over the I2R system in following this path. Overall, these results agree with our findings over the SemEval-2010 dataset (Section 3.2), un- derlining the viability of topic ... probability distribution over topics for each document, by simply aggregating the distributions over topics for each word in the document. In WSI terms, we take this distribu- tion over top...

Ngày tải lên: 17/03/2014, 22:20

... expressions (idioms) and other relations that hold over short ranges; larger window sizes will highlight semantic concepts and other relationships that hold over larger scales. For the remainder of this ... do keep. 2. Practical Applications The proposed statistical description has a large number of potentially important applications, including: (a) constraining the language mode...

Ngày tải lên: 24/03/2014, 02:20

... following: sim T T ∑ p i P T f i max p j P T alignment p i p j (1) sim T T is not commutative. That is, sim T T sim T T . alignment p i p j is an alignment score between the sequences p i and ... automatically. 3 Pairwise Alignment We attempt to apply the method of pairwise align- ment to measuring the similarity between sequences. Recently, the technique of pairwise alignment is wo...

Ngày tải lên: 08/03/2014, 04:22

... translation, significant improvements over IBM Model 4 in both word alignment (up to +6.7 F1) and translation quality (up to +1.4 Bleu). 1 Introduction Automatic word alignment is a vital component ... Expectation -Maximization (EM) algorithm (Dempster et al., 1977). 2.2 MAP-EM with the  0 -norm Maximum likelihood training is prone to over tting, especially in models with many paramet...

Ngày tải lên: 30/03/2014, 17:20

... parameters. We induced embeddings with 100 dimensions over 5-gram windows, and embeddings with 50 dimensions over 5-gram win- dows. Embeddings were induced over one pass 2 A rare word will appear 5 (window ... presented a neural language model that could be trained over billions of words, because the gradient of the loss was computed stochastically over a small sample of possible ou...

Ngày tải lên: 20/02/2014, 04:20