Automatic induction of bilingual resources from aligned parallel corpora: application to shallow-transfer machine translation

The availability of machine-readable bilingual linguistic resources is crucial not only for rule-based machine translation but also for other applications such as cross-lingual information retrieval. However, the building of such resources (bilingual single-word and multi-word correspondences, translation rules) demands extensive manual work, and, as a consequence, bilingual resources are usually more difficult to find than “shallow” monolingual resources such as morphological dictionaries or part-of-speech taggers, especially when they involve a less-resourced language. This paper describes a methodology to build automatically both bilingual dictionaries and shallow-transfer rules by extracting knowledge from word-aligned parallel corpora processed with shallow monolingual resources (morphological analysers, and part-of-speech taggers). We present experiments for Brazilian Portuguese–Spanish and Brazilian Portuguese–English parallel texts. The results show that the proposed methodology can enable the rapid creation of valuable computational resources (bilingual dictionaries and shallow-transfer rules) for machine translation and other natural language processing tasks).     

Unsupervised statistical text simplification using pre-trained language modeling for initialization

Unsupervised text simplification has attracted much attention due to the scarcity of high-quality parallel text simplification corpora. Recent an unsupervised statistical text simplification based on phrase-based machine translation system (UnsupPBMT) achieved good performance, which initializes the phrase tables using the similar words obtained by word embedding modeling. Since word embedding modeling only considers the relevance between words, the phrase table in UnsupPBMT contains a lot of dissimilar words. In this paper, we propose an unsupervised statistical text simplification using pre-trained language modeling BERT for initialization. Specifically, we use BERT as a general linguistic knowledge base for predicting similar words. Experimental results show that our method outperforms the state-of-the-art unsupervised text simplification methods on three benchmarks, even outperforms some supervised baselines.     

Building language resources for a Multi-Engine English-Filipino machine translation system

In this paper, we present the building of various language resources for a multi-engine bi-directional English-Filipino Machine Translation (MT) system. Since linguistics information on Philippine languages are available, but as of yet, the focus has been on theoretical linguistics and little is done on the computational aspects of these languages, attempts are reported here on the manual construction of these language resources such as the grammar, lexicon, morphological information, and the corpora which were literally built from almost non-existent digital forms. Due to the inherent difficulties of manual construction, we also discuss our experiments on various technologies for automatic extraction of these resources to handle the intricacies of the Filipino language, designed with the intention of using them for the MT system. To implement the different MT engines and to ensure the improvement of translation quality, other language tools (such as the morphological analyzer and generator, and the part of speech tagger) were developed.     

Tagging Icelandic text: an experiment with integrations and combinations of taggers

We use integrations and combinations of taggers to improve the tagging accuracy of Icelandic text. The accuracy of the best performing integrated tagger, which consists of our linguistic rule-based tagger for initial disambiguation and a trigram tagger for full disambiguation, is 91.80%. Combining five different taggers, using simple voting, results in 93.34% accuracy. By adding two linguistically motivated rules to the combined tagger, we obtain an accuracy of 93.48%. This method reduces the error rate by 20.5%, with respect to the best performing tagger in the combination pool.     

Matxin, an open-source rule-based machine translation system for Basque

We present the first publicly available machine translation (MT) system for Basque. The fact that Basque is both a morphologically rich and less-resourced language makes the use of statistical approaches difficult, and raises the need to develop a rule-based architecture which can be combined in the future with statistical techniques. The MT architecture proposed reuses several open-source tools and is based on a unique XML format to facilitate the flow between the different modules, which eases the interaction among different developers of tools and resources. The result is the rule-based Matxin MT system, an open-source toolkit, whose first implementation translates from Spanish to Basque. We have performed innovative work on the following tasks: construction of a dependency analyser for Spanish, use of rich linguistic information to translate prepositions and syntactic functions (such as subject and object markers), construction of an efficient module for verbal chunk transfer, and design and implementation of modules for ordering words and phrases, independently of the source language.     

融合EMD最小化双语词典的汉—越无监督神经机器翻译

神经机器翻译在平行语料充足的任务中能取得很好的效果,然而对于资源稀缺型语种的翻译任务则往往效果不佳.汉语和越南语之间没有大规模的平行语料库,在这项翻译任务中,该文探索只使用容易获得的汉语和越南语单语语料,通过挖掘单语语料中词级别的跨语言信息,融合到无监督翻译模型中提升翻译性能;该文提出了融合EMD(Earth Move...     

Corpus based part-of-speech tagging

In natural language processing, a crucial subsystem in a wide range of applications is a part-of-speech (POS) tagger, which labels (or classifies) unannotated words of natural language with POS labels corresponding to categories such as noun, verb or adjective. Mainstream approaches are generally corpus-based: a POS tagger learns from a corpus of pre-annotated data how to correctly tag unlabeled data. Presented here is a brief state-of-the-art account on POS tagging. POS tagging approaches make use of labeled corpus to train computational trained models. Several typical models of three kings of tagging are introduced in this article: rule-based tagging, statistical approaches and evolution algorithms. The advantages and the pitfalls of each typical tagging are discussed and analyzed. Some rule-based and stochastic methods have been successfully achieved accuracies of 93–96 %, while that of some evolution algorithms are about 96–97 %.     

Part-of-speech tagging of building codes empowered by deep learning and transformational rules

Automated building code compliance checking systems were under development for many years. However, the excessive amount of human inputs needed to convert building codes from natural language to computer understandable formats severely limited their range of applicable code requirements. To address that, automated code compliance checking systems need to enable an automated regulatory rule conversion. Accurate Part-of-Speech (POS) tagging of building code texts is crucial to this conversion. Previous experiments showed that the state-of-the-art generic POS taggers do not perform well on building codes. In view of that, the authors are proposing a new POS tagger tailored to building codes. It utilizes deep learning neural network model and error-driven transformational rules. The neural network model contains a pre-trained model and one or more trainable neural layers. The pre-trained model was fine-tuned on Part-of-Speech Tagged Building Codes (PTBC), a POS tagged building codes dataset. The fine-tuning of pre-trained model allows the proposed POS tagger to reach high precision with a small amount of available training data. Error-driven transformational rules were used to boost performance further by fixing errors made by the neural network model in the tagged building code. Through experimental testing, the authors found a well-performing POS tagger for building codes that had one bi-directional LSTM trainable layer, utilized BERT_Cased_Base pre-trained model and was trained 50 epochs. This model reached a 91.89% precision without error-driven transformational rules and a 95.11% precision with error-driven transformational rules, which outperformed the 89.82% precision achieved by the state-of-the-art POS taggers.     

跨语言词向量研究综述

随着人们对互联网多语言信息需求的日益增长,跨语言词向量已成为一项重要的基础工具,并成功应用到机器翻译、信息检索、文本情感分析等自然语言处理领域。跨语言词向量是单语词向量的一种自然扩展,词的跨语言表示通过将不同的语言映射到一个共享的低维向量空间,在不同语言间进行知识转移,从而在多语言环境下对词义进行准确捕捉。近几年跨语言词向量模型的研究成果比较丰富,研究者们提出了较多生成跨语言词向量的方法。该文通过对现有的跨语言词向量模型研究的文献回顾,综合论述了近年来跨语言词向量模型、方法、技术的发展。按照词向量训练方法的不同,将其分为有监督学习、无监督学习和半监督学习三类方法,并对各类训练方法的原理和代表性研究进行总结以及详细的比较;最后概述了跨语言词向量的评估及应用,并分析了所面临的挑战和未来的发展方向。     

A rule triggering system for automatic text-to-sign translation

The topic of this paper is machine translation (MT) from French text into French sign language (LSF). After arguing in favour of a rule-based method, it presents the architecture of an original MT system, built on two distinct efforts: formalising LSF production rules and triggering them with text processing. The former is made without any concern for text or translation and involves corpus analysis to link LSF form features to linguistic functions. It produces a set of production rules which may constitute a full LSF production grammar. The latter is an information extraction task from text, broken down in as many subtasks as there are rules in the grammar. After discussing this architecture, comparing it to the traditional methods and presenting the methodology for each task, the paper present the set of production rules found to govern event precedence and duration in LSF and gives a progress report on the implementation of the rule triggering system. With this proposal, it is also hoped to show how MT can benefit today from sign language processing.     

What types of word alignment improve statistical machine translation?

In most statistical machine translation (SMT) systems, bilingual segments are extracted via word alignment. However, there is a need for systematic study as to what alignment characteristics can benefit MT under specific experimental settings such as the type of MT system, the language pair or the type or size of the corpus. In this paper we perform, in each of these experimental settings, a statistical analysis of the data and study the sample correlation coefficients between a number of alignment or phrase table characteristics and variables such as the phrase table size, the number of untranslated words or the BLEU score. We report results for two different SMT systems (a phrase-based and an n-gram-based system) on Chinese-to-English FBIS and BTEC data, and Spanish-to-English European Parliament data. We find that the alignment characteristics which help in translation greatly depend on the MT system and on the corpus size. We give alignment hints to improve BLEU score, depending on the SMT system used and the type of corpus. For example, for phrase-based SMT, dense alignments are required with larger corpora, especially on the target side, while with smaller corpora, more precise, sparser alignments are better, especially on the source side. Avoiding some long-distance crossing links may also improve BLEU score with small corpora. We take these conclusions into account to modify two types of alignment systems, and get 1 to 1.6 % relative improvements in BLEU score on two held-out corpora, although the improved system is different in each corpus.     

Lightly supervised word-sense translation-error detection and resolution in an interactive conversational spoken language translation system

Lexical ambiguity can cause critical failure in conversational spoken language translation (CSLT) systems that rely on statistical machine translation (SMT) if the wrong sense is presented in the target language. Interactive CSLT systems offer the capability to detect and pre-empt such word-sense translation errors (WSTEs) by engaging the human operators in a precise clarification dialogue aimed at resolving the problem. This paper presents an end-to-end framework for accurate detection and interactive resolution of WSTEs to minimize communication errors due to ambiguous source words. We propose (a) a novel, extensible, two-level classification architecture for identifying potential WSTEs in SMT hypotheses; (b) a constrained phrase-pair clustering mechanism for identifying the translated sense of ambiguous source words in SMT hypotheses; and (c) an interactive strategy that integrates this information to request specific clarifying information from the operator. By leveraging unsupervised and lightly supervised learning techniques, our approach minimizes the need for expensive human annotation in developing each component of this framework. Each component, as well as the overall framework, was evaluated in the context of an interactive English-to-Iraqi Arabic CSLT system.     

Learning texture discrimination rules in a multiresolution system

We describe a texture analysis system in which informative discrimination rules are learned from a multiresolution representation of time textured input. The system incorporates unsupervised and supervised learning via statistical machine learning and rule-based neural networks, respectively. The textured input is represented in the frequency-orientation space via a log-Gabor pyramidal decomposition. In the unsupervised learning stage a statistical clustering scheme is used for the quantization of the feature-vector attributes. A supervised stage follows in which labeling of the textured map is achieved using a rule-based network. Simulation results for the texture classification task are given. An application of the system to real-world problems is demonstrated     

Multilingual and cross-domain temporal tagging

Extraction and normalization of temporal expressions from documents are important steps towards deep text understanding and a prerequisite for many NLP tasks such as information extraction, question answering, and document summarization. There are different ways to express (the same) temporal information in documents. However, after identifying temporal expressions, they can be normalized according to some standard format. This allows the usage of temporal information in a term- and language-independent way. In this paper, we describe the challenges of temporal tagging in different domains, give an overview of existing annotated corpora, and survey existing approaches for temporal tagging. Finally, we present our publicly available temporal tagger HeidelTime, which is easily extensible to further languages due to its strict separation of source code and language resources like patterns and rules. We present a broad evaluation on multiple languages and domains on existing corpora as well as on a newly created corpus for a language/domain combination for which no annotated corpus has been available so far.     

Generating Arabic text in multilingual speech-to-speech machine translation framework

The interlingual approach to machine translation (MT) is used successfully in multilingual translation. It aims to achieve the translation task in two independent steps. First, meanings of the source-language sentences are represented in an intermediate language-independent (Interlingua) representation. Then, sentences of the target language are generated from those meaning representations. Arabic natural language processing in general is still underdeveloped and Arabic natural language generation (NLG) is even less developed. In particular, Arabic NLG from Interlinguas was only investigated using template-based approaches. Moreover, tools used for other languages are not easily adaptable to Arabic due to the language complexity at both the morphological and syntactic levels. In this paper, we describe a rule-based generation approach for task-oriented Interlingua-based spoken dialogue that transforms a relatively shallow semantic interlingual representation, called interchange format (IF), into Arabic text that corresponds to the intentions underlying the speaker’s utterances. This approach addresses the handling of the problems of Arabic syntactic structure determination, and Arabic morphological and syntactic generation within the Interlingual MT approach. The generation approach is developed primarily within the framework of the NESPOLE! (NEgotiating through SPOken Language in E-commerce) multilingual speech-to-speech MT project. The IF-to-Arabic generator is implemented in SICStus Prolog. We conducted evaluation experiments using the input and output from the English analyzer that was developed by the NESPOLE! team at Carnegie Mellon University. The results of these experiments were promising and confirmed the ability of the rule-based approach in generating Arabic translation from the Interlingua taken from the travel and tourism domain.     

Part-of-speech tagger for Ainu language based on higher order Hidden Markov Model

This paper presents POST-AL, the first part-of-speech tagger for Ainu language. The system uses a hand-crafted dictionary based on Ainu narratives “yukar”. The system provides three types of information: word/token, part of speech, and translation of the token (in Japanese). Evaluation on a training set provided positive results. The system could be useful in a great number of tasks related to the research on Ainu language, such as content analysis or translation, which till now have been done mostly manually.     

基于知网义原词向量表示的无监督词义消歧方法

词义消歧一直是自然语言处理领域中的重要问题,该文将知网(HowNet)中表示词语语义的义原信息融入到语言模型的训练中。通过义原向量对词语进行向量化表示,实现了词语语义特征的自动学习,提高了特征学习效率。针对多义词的语义消歧,该文将多义词的上下文作为特征,形成特征向量,通过计算多义词词向量与特征向量之间相似度进行词语消歧。作为一种无监督的方法,该方法大大降低了词义消歧的计算和时间成本。在SENSEVAL-3的测试数据中准确率达到了37.7%,略高于相同测试集下其他无监督词义消歧方法的准确率。     

一种混合策略的专利机器翻译系统研究

针对专利文本翻译中的复杂语句,提出了一种基于混合策略的方法,融合语义分析技术和基于规则的翻译技术,来提高专利翻译的效果。利用语义分析技术,重点解决句子中心动词识别和句子中有嵌套结构存在的名称短语的分析,把语义分析结果输入到基于规则的翻译系统中,用以改善翻译的效果。测试结果表明,融合后的翻译系统,BLEU值提高了9.8%。该方法已经集成到了国家知识产权局的在线汉英机器翻译系统中,有效地提高了专利翻译的效果和翻译效率。     

基于变分信息瓶颈的半监督神经机器翻译

变分方法是机器翻译领域的有效方法, 其性能较依赖于数据量规模. 然而在低资源环境下, 平行语料资源匮乏, 不能满足变分方法对数据量的需求, 因此导致基于变分的模型翻译效果并不理想. 针对该问题, 本文提出基于变分信息瓶颈的半监督神经机器翻译方法, 所提方法的具体思路为: 首先在小规模平行语料的基础上, 通过引入跨层注意力机制充分利用神经网络各层特征信息, 训练得到基础翻译模型; 随后, 利用基础翻译模型, 使用回译方法从单语语料生成含噪声的大规模伪平行语料, 对两种平行语料进行合并形成组合语料, 使其在规模上能够满足变分方法对数据量的需求; 最后, 为了减少组合语料中的噪声, 利用变分信息瓶颈方法在源与目标之间添加中间表征, 通过训练使该表征具有放行重要信息、阻止非重要信息流过的能力, 从而达到去除噪声的效果. 多个数据集上的实验结果表明, 本文所提方法能够显著地提高译文质量, 是一种适用于低资源场景的半监督神经机器翻译方法.