Pasar al contenido principal

Resultados relacionados

Resultados totales (Incluyendo duplicados): 2
Encontrada(s) 1 página(s)

RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

Online learning for interactive statistical machine translation

  • Ortiz-Martínez, Daniel
  • García-Varea, Ismael
  • Casacuberta Nolla, Francisco|||0000-0002-8497-5598
[EN] State-of-the-art Machine Translation (MT) systems are still far from being perfect. An alternative is the so-called Interactive Machine Translation (IMT) framework. In this framework, the knowledge of a human translator is combined with a MT system. The vast majority of the existing work on IMT makes use of the well-known batch learning paradigm. In the batch learning paradigm, the training of the IMT system and the interactive translation process are carried out in separate stages. This paradigm is not able to take advantage of the new knowledge produced by the user of the IMT system. In this paper, we present an application of the online learning paradigm to the IMT framework. In the online learning paradigm, the training and prediction stages are no longer separated. This feature is particularly useful in IMT since it allows the user feedback to be taken into account. The online learning techniques proposed here incrementally update the statistical models involved in the translation process. Empirical results show the great potential of online learning in the IMT framework., Work supported by the EC (FEDER/FSE), the Spanish Government (MEC, MICINN, MITyC, MAEC, ¿Plan E¿, under grants MIPRCV ¿Consolider Ingenio 2010¿ CSD2007-00018, iTrans2 TIN2009-14511, erudito.com TSI-020110-2009-439), the Generalitat Valenciana (grant Prometeo/2009/014), the Univ. Politecnica de Valencia (grant 20091027) and the Spanish JCCM (grant PBI08-0210-7127).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

Does more data always yield better translations?

  • Gascó Mora, Guillem
  • Rocha Sánchez, Martha Alicia
  • Sanchis Trilles, Germán
  • Andrés Ferrer, Jesús
  • Casacuberta Nolla, Francisco|||0000-0002-8497-5598
Nowadays, there are large amounts of data available to train statistical machine translation systems. However, it is not clear whether all the training data actually help or not. A system trained on a subset of such huge bilingual corpora might outperform the use of all the bilingual data. This paper studies such issues by analysing two training data selection techniques: one based on approximating the probability of an indomain corpus; and another based on infrequent n-gram occurrence. Experimental results not only report significant improvements over random sentence selection but also an improvement over a system trained with the whole available data. Surprisingly, the improvements are obtained with just a small fraction of the data that accounts for less than 0.5% of the sentences. Afterwards, we show that a much larger room for improvement exists, although this is done under non-realistic conditions., The research leading to these results has received funding from the European Union Seventh Framework Programme (FP7/2007-2013) under grant agreement nr. 287755. This work was also supported by the Spanish MEC/MICINN under the MIPRCV ”Consolider Ingenio 2010” program (CSD2007-00018), and iTrans2 (TIN2009-14511) project. Also supported by the Spanish MITyC under the erudito.com (TSI-020110-2009-439) project and Instituto Tecnológico de León, DGEST-PROMEP y CONACYT, México.