Detecting Document-level Context Triggers to Resolve Translation Ambiguity

Abstract

Most current machine translation systems translate each sentence independently, ignoring the context from previous sentences. This discourse unawareness can lead to incorrect translation of words or phrases that are ambiguous in the sentence. For example, the German term Typen in the phrase diese Typen can be translated either into English types or guys. However, knowing that it co-refers to the compound Körpertypen ("body types") in the previous sentence helps to disambiguate the term and translate it into types. We propose a method of automatically detecting document-level trigger words (like Körpertypen), whose presence helps to disambiguate translations of ambiguous terms. In this preliminary study we analyze the method and its limitations, and outline future work directions.

On Integrating Discourse in Machine Translation

“…In their work, Mascarell et al (2015) use trigger words from the ST to try to disambiguate translations of ambiguous terms, where a word in the source language can have different meanings and should be rendered with a different lexical item in the TT depending on the context it occurs in.…”

“…Despite the fact that discourse has long been recognised as a crucial part of translation (Hatim and Mason, 1990), when it comes to Statistical Machine Translation (SMT), discourse information has been mostly neglected. Recently increasing amounts of effort have been going into addressing discourse explicitly in MT, with research covering lexical cohesion (Wong and Kit, 2012;Xiong et al, 2013b,a;Gong et al, 2015;Mascarell et al, 2015), discourse connectives (Cartoni et al, 2012(Cartoni et al, , 2013Meyer, 2011;Steele, 2015;Steele and Specia, 2016), discourse relations (Guzmán et al, 2014), pronoun prediction (Guillou, 2012;Hardmeier et al, 2013b;Guillou, 2016) and negation (Fancellu and Webber, 2014;Wetzel and Bond, 2012).…”