Improving Interactive Machine Translation via Mouse Actions

Proceedings of the 2008 Conference on Empirical Methods in Natural Language Processing, pages 485–494, Honolulu, October 2008. c⃝2008 Association for Computational Linguistics Improving Interactive Machine Translation via Mouse Actions Germ´an Sanchis-Trilles and Daniel Ortiz-Mart´ınez and Jorge Civera Instituto Tecnol´ogico de Inform´atica Universidad Polit´ecnica de Valencia {gsanchis,dortiz,jorcisai}@iti.upv.es Francisco Casacuberta and Enrique Vidal Departamento de Sistemas Inform´aticos y Computaci´on Universidad Polit´ecnica de Valencia {fcn,evidal}@dsic.upv.es Hieu Hoang University of Edinburgh hhoang@sms.ed.ac.uk Abstract Although Machine Translation (MT) is a very active research field which is receiving an in- creasing amount of attention from the research community, the results that current MT sys- tems are capable of producing are still quite far away from perfection. Because of this, and in order to build systems that yield correct translations, human knowledge must be inte- grated into the translation process, which will be carried out in our case in an Interactive- Predictive (IP) framework. In this paper, we show that considering Mouse Actions as a sig- nificant information source for the underly- ing system improves the productivity of the human translator involved. In addition, we also show that the initial translations that the MT system provides can be quickly improved by an expert by only performing additional Mouse Actions. In this work, we will be using word graphs as an efficient interface between a phrase-based MT system and the IP engine. 1 Introduction Information technology advances in modern society have led to the need of more efficient methods of translation. It is important to remark that current MT systems are not able to produce ready-to-use texts (Kay, 1997; Hutchins, 1999; Arnold, 2003). Indeed, MT systems are usually limited to specific semantic domains and the translations provided re- quire human post-editing in order to achieve a cor- rect high-quality translation. A way of taking advantage of MT systems is to combine them with the knowledge of a human trans- lator, constituting the so-called Computer-Assisted Translation (CAT) paradigm. CAT offers different approaches in order to benefit from the synergy be- tween humans and MT systems. An important contribution to interactive CAT technology was carried out around the TransType (TT) project (Langlais et al., 2002; Foster et al., 2002; Foster, 2002; Och et al., 2003). This project entailed an interesting focus shift in which interac- tion directly aimed at the production of the target text, rather than at the disambiguation of the source text, as in former interactive systems. The idea proposed was to embed data driven MT techniques within the interactive translation environment. Following these TT ideas, (Barrachina and oth- ers, 2008) propose the usage of fully-fledged statis- tical MT (SMT) systems to produce full target sen- tence hypotheses, or portions 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 infor- mation to achieve further, hopefully improved sug- gestions. In this paper, we also focus on the inter- active and predictive, statistical MT (IMT) approach to CAT. The IMT paradigm …

Content truncated to 500 words