EXTRA: a system for example-based translation assistance

Abstract

In this paper we present EXTRA (EXample-based TRanslation Assistant), a translation memory (TM) system. EXTRA is able to propose effective translation suggestions by relying on syntactic analysis of the text and on a rigorous, language-independent measure; the search is performed efficiently in large amounts of bilingual texts thanks to its advanced retrieval techniques. EXTRA does not use external knowledge requiring the intervention of users and is completely customizable and portable as it has been implemented on top of a standard DataBase Management System. The paper provides a thorough evaluation of both the effectiveness and the efficiency of our system. In particular, in order to quantify the benefits offered by EXTRA assisted translation over manual translation, we introduce a simulator implementing specifically devised statistical, process-oriented, discrete-event models. As far as we know, this is the first time statistical simulation experiments have been used to face the nontrivial problem of evaluating TM systems, particularly for comparing the time that could be saved by performing assisted translation versus "manual" translation and for optimally tuning the system behaviour with respect to differently skilled users. In our experiments, we considered three scenarios, manual translation with one or two translators and assisted translation with one translator. The time needed for one translator to do an assisted translation is significantly closer to that of a team of two translators than to that of the single translator. The mean sentence translation time is by far the lowest for this scenario, corresponding to the highest per translator productivity. 123 168 F. Mandreoli et al.We also estimate the total translation time when the number of query sentences, the maximum number of suggestions to be read, and the probability of look up are varied: the best trade-off is given by reading (and presenting) four or five suggestions at the most.

Translation memory retrieval methods

“…The retrieval algorithm used by commercial TM systems is typically not disclosed (Koehn and Senellart, 2010;Simard and Fujita, 2012;Whyman and Somers, 1999). However, the bestperforming method used in current systems is widely believed to be based on edit distance (Baldwin and Tanaka, 2000;Simard and Fujita, 2012;Whyman and Somers, 1999;Koehn and Senellart, 2010;Christensen and Schjoldager, 2010;Mandreoli et al, 2006;He et al, 2010). Recently Simard and Fujita (2012) have experimented with using MT (machine translation) evaluation metrics as TM fuzzy match, or similarity, algorithms.…”

Translation-Memory (TM) Research: What Do We Know and How Do We Know It?

“…Our analysis suggests that empirical TM research may be divided roughly into three areas according to authors' interests and orientation: Mandreoli et al (2006) and Colominas (2008) seem to be interested mainly in the (1) technical side of TM-assisted translation. This kind of empirical research may be referred to as technology-oriented.…”

“…All studies either take for granted (Mandreoli et al 2006;Colominas 2008;Lange/Bennett 2000;Christensen 2003;Dragsted 2004Dragsted , 2006O'Brien 2006O'Brien , 2008 or document (Fulford/ Granell-Zafra 2005;Dillon/Fraser 2006;Lagoudaki 2006) that TM technology is here for good, as all types of professional translators seem to be aware of it and many -though not all -make good use of it.…”

“…Mandreoli et al (2006) and Colominas (2008) base their fi ndings on technical test runs. Christensen (2003) combines technical test runs with analyses of textual corpora.…”

“…A rather surprising result is that, though they own a TM tool, some respondents (16%) have not been able to learn how to use it yet. Mandreoli et al (2006) evaluate the design of their own TM, namely the Example-based TRanslation Assistant (EXTRA). In comparison with most TM systems relying on artifi cial intelligence, the search engine of the EXTRA system is founded on advanced information retrieval techniques executing two processes: fi rst, the document to be translated undergoes a syntactic analysis, and then it is compared with the TM data using so-called edit distance.…”