Touch-Based Pre-Post-Editing of Machine Translation Output

Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, pages 1040–1045, Lisbon, Portugal, 17-21 September 2015. c⃝2015 Association for Computational Linguistics. Touch-Based Pre-Post-Editing of Machine Translation Output Benjamin Marie LIMSI-CNRS, Orsay, France Lingua et Machina, Le Chesnay, France benjamin.marie@limsi.fr Aur´elien Max LIMSI-CNRS, Orsay, France Univ. Paris Sud, Orsay, France aurelien.max@limsi.fr Abstract We introduce pre-post-editing, possibly the most basic form of interactive trans- lation, as a touch-based interaction with iteratively improved translation hypothe- ses prior to classical post-editing. We re- port simulated experiments that yield very large improvements on classical evalua- tion metrics (up to 21 BLEU) as well as on a parameterized variant of the TER metric that takes into account the cost of matching/touching tokens, confirming the promising prospects of the novel transla- tion scenarios offered by our approach. 1 Introduction As shown by oracle studies (Wisniewski et al., 2010; Turchi et al., 2012; Marie and Max, 2013), Statistical Machine Translation (SMT) systems produce results that are of significantly lower qual- ity than what could be produced from their avail- able resources. As a pragmatic solution, human intervention is commonly used for improving au- tomatic draft translations, in so-called post-editing (PE), but is also studied earlier in the translation process in a variety of interactive strategies, in- cluding e.g. completion assistance and local trans- lation choices (e.g. (Foster et al., 2002; Koehn and Haddow, 2009; Gonz´alez-Rubio et al., 2013)). Al- though interactive machine translation does facil- itate the work of the SMT system in certain situa- tions by allowing it to make efficient use of knowl- edge contributed by the human translator, post- editing has been shown to remain a faster alter- native (Green et al., 2014). Nevertheless, this ac- tivity usually requires complex intervention from an expert translator (Carl et al., 2011). In this work we reduce interaction with an SMT system to its most basic form: similarly to what a human translator is likely to do when first reading a draft translation to post-edit, we require a user to simply spot those segments of a draft transla- tion that can participate in an acceptable transla- tion. The corresponding information is then used by a SMT system in a soft way to improve the draft translation. This process may be iteratively repeated as long as enough improvements are ob- tained, and terminates with classical post-editing on the obtained translation, hence we dub it pre- post-editing (PPE). We resort to simulated pre- post-editing and post-editing, as in other works (Carl et al., 2011; Denkowski et al., 2014), to measure translation performance on some avail- able reference translation using both classical met- rics and a variant of the TER metric (Snover et al., 2006), where, essentially, the cost of a token matching operation is a parameterized fraction of the cost of the other token edit operations. With the implementation of appropriate strategies in the SMT system, we show under reasonable assump- tions that this approach has the potential to signifi- cantly reduce the amount of human effort …

Content truncated to 500 words