Search

In this paper, we report our experiments in combining two EBMT systems that rely on generalized templates, Marclator and CMU-EBMT, on an English–German translation task. Our goal was to see whether a statistically significant improvement could be achieved over the individual performances of these two systems. We observed that this was not the case. However, our system consistently outperformed a lexical EBMT baseline system.

This paper reports MT evaluation experiments that were conducted at the end of year 1 of the EU-funded CoSyne 1 project for three language combinations, considering translations from German, Italian and Dutch into English. We present a comparative evaluation of the MT software developed within the project against four of the leading free webbased MT systems across a range of state-of-the-art automatic evaluation metrics. The data sets from the news domain that were created and used for training purposes and also for this evaluation exercise, which are available to the research community, are also described. The evaluation results for the news domain are very encouraging: the CoSyne MT software consistently beats the rule-based MT systems, and for translations from Italian and Dutch into English in particular the scores given by some of the standard automatic evaluation metrics are not too distant from those obtained by wellestablished statistical online MT systems.

This paper presents a set of experiments on Domain Adaptation of Statistical Machine Translation systems. The experiments focus on Chinese-English and two domain-specific corpora. The paper presents a novel approach for combining multiple domain-trained translation models to achieve improved translation quality for both domain-specific as well as combined sets of sentences. We train a statistical classifier to classify sentences according to the appropriate domain and utilize the corresponding domain-specific MT models to translate them. Experimental results show that the method achieves a statistically significant absolute improvement of 1.58 BLEU (2.86% relative improvement) score over a translation model trained on combined data, and considerable improvements over a model using multiple decoding paths of the Moses decoder, for the combined domain test set. Furthermore, even for domain-specific test sets, our approach works almost as well as dedicated domain-specific models and perfect classification.

Recent research has shown that Phrase-Based Statistical Machine Translation (PB-SMT) systems can benefit from two enhancements: (i) using words and POS tags as context-informed features on the source side; and (ii) incorporating lexical syntactic descriptions in the form of supertags on the target side. In this work we present a novel PB-SMT model that combines these two aspects by using supertags as source language contextinformed features. These features enable us to exploit source similarity in addition to target similarity, as modelled by the language model. In our experiments two kinds of supertags are employed: those from Lexicalized Tree-Adjoining Grammar and Combinatory Categorial Grammar. We use a memory-based classification framework that enables the estimation of these features while avoiding problems of sparseness. Despite the differences between these two approaches, the supertaggers give similar improvements. We evaluate the performance of our approach on an English-to-Chinese translation task using a state-of-the-art phrase-based SMT system, and report an improvement of 7.88% BLEU score in translation quality when adding supertags as context-informed features.

The Phrase-Based Statistical Machine Translation (PB-SMT) model has recently begun to include source context modeling, under the assumption that the proper lexical choice of an ambiguous word can be determined from the context in which it appears. Various types of lexical and syntactic features such as words, parts-of-speech, and supertags have been explored as effective source context in SMT. In this paper, we show that position-independent syntactic dependency relations of the head of a source phrase can be modeled as useful source context to improve target phrase selection and thereby improve overall performance of PB-SMT. On a Dutch—English translation task, by combining dependency relations and syntactic contextual features (part-of-speech), we achieved a 1.0 BLEU (Papineni et al., 2002) point improvement (3.1% relative) over the baseline.

Statistical Machine Translation (SMT) systems are usually trained on large amounts of bilingual text and monolingual target language text. If a significant amount of out-of-domain data is added to the training data, the quality of translation can drop. On the other hand, training an SMT system on a small amount of training material for given indomain data leads to narrow lexical coverage which again results in a low translation quality. In this paper, (i) we explore domain-adaptation techniques to combine large out-of-domain training data with small-scale in-domain training data for English—Hindi statistical machine translation and (ii) we cluster large out-of-domain training data to extract sentences similar to in-domain sentences and apply adaptation techniques to combine clustered sub-corpora with in-domain training data into a unified framework, achieving a 0.44 absolute corresponding to a 4.03% relative improvement in terms of BLEU over the baseline.

Statistical machine translation (SMT) models have recently begun to include source context modeling, under the assumption that the proper lexical choice of the translation for an ambiguous word can be determined from the context in which it appears. Various types of lexical and syntactic features have been explored as effective source context to improve phrase selection in SMT. In the present work, we introduce lexico-syntactic descriptions in the form of supertags as source-side context features in the state-of-the-art hierarchical phrase-based SMT (HPB) model. These features enable us to exploit source similarity in addition to target similarity, as modelled by the language model. In our experiments two kinds of supertags are employed: those from lexicalized tree-adjoining grammar (LTAG) and combinatory categorial grammar (CCG). We use a memory-based classification framework that enables the efficient estimation of these features. Despite the differences between the two supertagging approaches, they give similar improvements. We evaluate the performance of our approach on an English-to-Dutch translation task, and report statistically significant improvements of 4.48% and 6.3% BLEU scores in translation quality when adding CCG and LTAG supertags, respectively, as context-informed features.