Machine Translation (MT) for language pairs with long distance dependencies and word reordering, such as German-English, is prone to producing output that is lexically or syntactically incoherent. Statistical MT (SMT) models used explicit or latent syntax to improve reordering, however failed at capturing other long distance dependencies. This thesis explores how explicit sentence-level syntactic information can improve translation for such complex linguistic phenomena. In particular, we work at the level of the syntactic-semantic interface with representations conveying the predicate-argument structures. These are essential to preserving semantics in translation and SMT systems have long struggled to model them. String-to-tree SMT systems use explicit target syntax to handle long-distance reordering, but make strong independence assumptions which lead to inconsistent lexical choices. To address this, we propose a Selectional Preferences feature which models the semantic affinities between target predicates and their argument fillers using the target dependency relations available in the decoder. We found that our feature is not effective in a string-to-tree system for German-English and that often the conditioning context is wrong because of mistranslated verbs. To improve verb translation, we proposed a Neural Verb Lexicon Model (NVLM) incorporating sentence-level syntactic context from the source which carries relevant semantic information for verb disambiguation. When used as an extra feature for re-ranking the output of a German-English string-to-tree system, the NVLM improved verb translation precision by up to 2.7% and recall by up to 7.4%. While the NVLM improved some aspects of translation, other syntactic and lexical inconsistencies are not being addressed by a linear combination of independent models. In contrast to SMT, neural machine translation (NMT) avoids strong independence assumptions thus generating more fluent translations and capturing some long-distance dependencies. Still, incorporating additional linguistic information can improve translation quality. We proposed a method for tightly coupling target words and syntax in the NMT decoder. To represent syntax explicitly, we used CCG supertags, which encode subcategorization information, capturing long distance dependencies and attachments. Our method improved translation quality on several difficult linguistic constructs, including prepositional phrases which are the most frequent type of predicate arguments. These improvements over a strong baseline NMT system were consistent across two language pairs: 0.9 BLEU for German-English and 1.2 BLEU for Romanian-English.

Работа посвящена сравнению качества перевода технического текста с английского языка на русский, полученного с помощью различных систем машинного перевода (СМП), на примере научной статьи металлургической тематики “Fluid Dynamics Studies of Bottom-blown and Side-blown Copper Smelting Furnaces”. В первой части работы рассматриваются общие вопросы машинного и технического перевода: приводятся понятия машинного перевода и технического перевода, определяются основные способы и алгоритмы работы СМП, выделяются основные особенности и проблемы технического перевода. Особое внимание уделяется выявлению функциональных возможностей наиболее популярных сервисов машинного перевода, к которым относятся такие СМП как Google Translate, Яндекс Переводчик, Bing Microsoft Translator, SYSTRAN Translate, PROMT.One. Во второй части работы раскрываются особенности металлургической терминологии и специфика ее заимствования, разрабатывается методика оценки качества перевода технического текста металлургической направленности, проводится апробационное исследование переводов, выполненных с помощью различных систем машинного перевода и определяются перспективы использования СМП в металлургической отрасли. Большое внимание уделяется оценке качества переводов по представленным методикам, при этом все результаты приводятся в виде таблиц с указанием количества ошибок и итоговых баллов. Результаты исследования могут быть использованы на различных предприятиях металлургического сектора при работе с документацией на иностранных языках. / This paper is devoted to comparison of translation quality obtained as a result of using machine translation (MT) systems to translate technical documents from English into Russian in the case of a metallurgical article “Fluid Dynamics Studies of Bottom-blown and Side-blown Copper Smelting Furnaces”. The first part of the paper covers general issues of machine and technical translation, including concepts of machine and technical translation, the main operation methods and algorithms, and the main features of technical translation. Particular attention is paid to identification of core functionality of the most popular machine translation systems, which include Google Translate, Yandex Translate, Bing Microsoft Translator, SYSTRAN Translate, PROMT.One. The second part of the paper reveals the main features of metallurgical terminology and the specifics of its naturalization. It also contains several translation quality assessment methodologies and includes an analysis of machine translation quality. Moreover, it features the main prospects for using MT systems in metallurgy. Much attention is paid to translation quality assessment on methodologies presented in the paper. All the results are presented in the form of tables with the number of errors and final scores for each MT system. The results of the study may be used at various metallurgical enterprises working with documentation in foreign languages.