Online Learning for Interactive Statistical Machine Translation

Human Language Technologies: The 2010 Annual Conference of the North American Chapter of the ACL, pages 546–554, Los Angeles, California, June 2010. c⃝2010 Association for Computational Linguistics Online Learning for Interactive Statistical Machine Translation Daniel Ortiz-Mart´ınez Dpto. de Sist. Inf. y Comp. Univ. Polit´ec. de Valencia 46071 Valencia, Spain dortiz@dsic.upv.es Ismael Garc´ıa-Varea Dpto. de Inform´atica Univ. de Castilla-La Mancha 02071 Albacete, Spain ivarea@info-ab.uclm.es Francisco Casacuberta Dpto. de Sist. Inf. y Comp. Univ. Polit´ec. de Valencia 46071 Valencia, Spain fcn@dsic.upv.es Abstract State-of-the-art Machine Translation (MT) systems are still far from being perfect. An al- ternative is the so-called Interactive Machine Translation (IMT) framework. In this frame- work, the knowledge of a human translator is combined with a MT system. The vast ma- jority 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 ap- plication 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 par- ticularly useful in IMT since it allows the user feedback to be taken into account. The online learning techniques proposed here incremen- tally update the statistical models involved in the translation process. Empirical results show the great potential of online learning in the IMT framework. 1 Introduction Information technology advances have led to the need for more efficient translation methods. Current MT systems are not able to produce ready-to-use texts. Indeed, MT systems usually require human post-editing to achieve high-quality translations. One way of taking advantage of MT systems is to combine them with the knowledge of a human trans- lator in the IMT paradigm, which is a special type of the computer-assisted translation paradigm (Isabelle and Church, 1997). An important contribution to IMT technology was pioneered by the TransType project (Foster et al., 1997; Langlais et al., 2002) where data driven MT techniques were adapted for their use in an interactive translation environment. Following the TransType ideas, Barrachina et al. (2009) proposed a new approach to IMT, in which fully-fledged statistical MT (SMT) systems are used to produce full target sentence hypotheses, or por- tions thereof, which can be partially or completely accepted and amended by a human translator. Each partial, correct text segment is then used by the SMT system as additional information to achieve improved suggestions. Figure 1 illustrates a typical IMT session. source(f): Para ver la lista de recursos reference(ˆe): To view a listing of resources inter.-0 ep es To view the resources list inter.-1 ep To view k a es list of resources inter.-2 ep To view a list k list i es list i ng resources inter.-3 ep To view a listing k o es o f resources accept ep To …

Content truncated to 500 words