We reassess a recent study (Hassan et al., 2018) that claimed machine translation (MT) has reached human parity for the of news from Chinese into English, using pairwise ranking and considering three variables were not taken account in previous study: language which source side test set was originally written, proficiency evaluators, provision inter-sentential context. If we consider only original text (i.e. translated another language, or translationese), then find evidence showing been...

In translation, considering the document as a whole can help to resolve ambiguities and inconsistencies. this paper, we propose cross-sentence context-aware approach investigate influence of historical contextual information on performance neural machine translation (NMT). First, history is summarized in hierarchical way. We then integrate representation into NMT two strategies: 1) warm-start encoder decoder states, 2) an auxiliary context source for updating states. Experimental results...

Abstract This paper discusses neural machine translation (NMT), a new paradigm in the MT field, comparing quality of NMT systems with statistical by describing three studies using automatic and human evaluation methods. Automatic results presented for are very promising, however evaluations show mixed results. We report increases fluency but inconsistent adequacy post-editing effort. undoubtedly represents step forward one that community should be careful not to oversell.

Speakers of different languages must attend to and encode strikingly aspects the world in order use their language correctly (Sapir, 1921; Slobin, 1996). One such difference is related way gender expressed a language. Saying "I am happy" English, does not any additional knowledge speaker that uttered sentence. However, many other do have grammatical systems so would be encoded. In translate sentence into, say, French, inherent information needs retained/recovered. The same become either "Je...

A prerequisite for training corpus-based machine translation (MT) systems -- either Statistical MT (SMT) or Neural (NMT) is the availability of high-quality parallel data. This arguably more important today than ever before, as NMT has been shown in many studies to outperform SMT, but mostly when large corpora are available; cases where data limited, SMT can still NMT. Recently researchers have that back-translating monolingual be used create synthetic corpora, which turn combination with...

Consistency is a key requirement of high-quality translation. It especially important to adhere pre-approved terminology and adapt corrected translations in domain-specific projects. Machine translation (MT) has achieved significant progress the area domain adaptation. However, real-time adaptation remains challenging. Large-scale language models (LLMs) have recently shown interesting capabilities in-context learning, where they learn replicate certain input-output text generation patterns,...

example, where the machine side requires some standard expressions in order to execute certain actions.Language translation is one of most complicated tasks human brain, which utilizes not only linguistic knowledge but also world, and varieties our sophisticated senses.EBMT possible approaches mechanism translation, this book represents a considerable contribution field.But we have investigate many other possibilities approach level complex functions brain.

Abstract In the context of recent improvements in quality machine translation (MT) output and new use cases being found for that output, this article reports on an experiment using statistical neural MT systems to translate literature. Six professional translators with experience literary produced English-to-Catalan translations under three conditions: from scratch, post-editing, post-editing. They provided feedback before after via questionnaires interviews. While all participants prefer...

We conduct the first experiment in literature which a novel is translated automatically and then post-edited by professional literary translators. Our case study Warbreaker, popular fantasy originally written English, we translate into Catalan. one chapter of (over 3,700 words, 330 sentences) with two data-driven approaches to Machine Translation (MT): phrase-based statistical MT (PBMT) neural (NMT). Both systems are tailored novels; they trained on over 100 million words fiction. In post-...

Attention mechanism is often used in deep neural networks for distantly supervised relation extraction (DS-RE) to distinguish valid from noisy instances. However, traditional 1-D vector attention model insufficient learning of different contexts the selection instances predict relationship an entity pair. To alleviate this issue, we propose a novel multi-level structured (2-D matrix) self-attention DS-RE multi-instance (MIL) framework using bidirectional recurrent (BiRNN). In proposed...

This paper shows how finite approximations of long distance dependency (LDD) resolution can be obtained automatically for wide-coverage, robust, probabilistic Lexical-Functional Grammar (LFG) resources acquired from treebanks. We extract LFG subcategorisation frames and paths linking LDD reentrancies f-structures generated the Penn-II treebank trees use them in an algorithm to parse new text. Unlike (Collins, 1999; Johnson, 2000), our approach LDDs is done at f-structure (attribute-value...

Contrary to perceived wisdom, we explore the role of machine translation (MT) in assisting with literary texts, considering both its limitations and potential. Our motivations this subject are twofold, arising from: (1) recent research advances MT, (2) emergence ebook, which together allow us for first time build literature-specific MT systems by training statistical models on novels their professional translations. A key challenge is that one needs preserve not only meaning (as other...

Abstract This paper presents an overview of Statistical Machine Translation (SMT), which is currently the dominant approach in (MT) research. In Way and Hearne (2011) , we describe how central linguists translators are to MT process, so that SMT developers researchers may better understand include these groups continuing advance state‐of‐the‐art. If constituencies make impact field MT, they need know their input used by systems. Accordingly, our objective this present basic principles...

Longyue Wang, Zhaopeng Tu, Xiaojun Zhang, Hang Li, Andy Way, Qun Liu. Proceedings of the 2016 Conference North American Chapter Association for Computational Linguistics: Human Language Technologies. 2016.

This work presents an empirical approach to quantifying the loss of lexical richness in Machine Translation (MT) systems compared Human (HT). Our experiments show how current MT indeed fail render diversity human generated or translated text. The inability generate diverse outputs and its tendency exacerbate already frequent patterns while ignoring less ones, might be underlying cause for, among others, currently heavily debated issues related gender biased output. Can we indeed, aside from...