A Computer-Assisted Translation Tool based on Finite-State Technology

A Computer-Assisted Translation Tool based on Finite-State Technology∗ Jorge Civera♣, Juan M. Vilar♦, Antonio L. Lagarda♣, Elsa Cubel♠, Sergio Barrachina♥, Francisco Casacuberta,♣, Enrique Vidal♠ ♣Departamento de Sistemas Inform´aticos y Computaci´on. Univ. Polit`ecnica de Val`encia (Spain) ♠Instituto Tecnol´ogico de Inform´atica. Univ. Polit`ecnica de Val`encia (Spain) ♦Departamento de Lenguajes y Sistemas Inform´aticos. Univ. Jaume I (Spain) ♥Departamento de Ingeniera y Ciencia de Computadores. Univ. Jaume I (Spain) tt2iti@iti.upv.es Abstract The Computer-Assisted Translation (CAT) paradigm tries to integrate hu- man expertise into the automatic translation process. In this paradigm, a human translator interacts with a translation system that dynamically offers a list of translations that best completes the part of the sentence that is being translated. This human-machine sinergy aims at a double goal, to increase translator productivity and ease translators’ work. In this paper, we present a CAT system based on stochastic finite-state transducer tech- nology. This system has been developed and assessed on two real parallel corpora in the framework of the European project TransType2 (TT2). 1 Introduction Information technology advances in modern society have led to the need of more efficient methods of translation. It is important to emphasise that current Machine Translation (MT) systems are not able to produce ready- to-use text. Indeed, MT systems are usually limited to specific semantic domains and the translations provided require post-editing in order to achieve a correct translation. A way of taking advantage of MT sys- tems is to combine them with the knowl- edge of a human translator by means of a Computer-Assisted Translation (CAT) sys- tem. In a CAT system, the user can amend the translation offered by the MT system, as the system takes into account these cor- rections to improve its translation. In this type of systems, the translator can work in a more comfortable way because of the greater freedom to make changes at any time while the translation is in progress. ∗This work has been supported by the European Union under the IST Programme (IST-2001-32091) and the Spanish project TIC2003-08681-C02. The CAT approach has two important as- pects: the models need to provide adequate completions and they have to do so effi- ciently under usability constrains. To fulfill these two requirements, Stochastic Finite- State Transducers (SFST) have been se- lected since they have proved to be able to provide adequate translations (Knight & Al-Onaizan, 1998; Amengual et al., 2000; Bangalore & Riccardi, 1995). In addition, efficient parsing algorithms can be easily adapted in order to provide completions. The rest of the paper is structured as fol- lows. Next section introduces the general setting for MT and finite-state models. In Section 3, the search procedure for interac- tive translation is explained. Experimental results are presented in Section 4. Finally, some conclusions and future work are ex- posed in Section 5. 2 Machine translation with finite-state transducers In a probabilistic framework, given a source sentence s, the goal of MT is to find a target sentence ˆt that: ˆt=argmax t Pr(t | s) = argmax t Pr(t, s). …

Content truncated to 500 words