Small: RI: Broad-Coverage High-Accuracy Machine Translation into Morphologically-Rich Languages

机器翻译（MT）到形态丰富的语言提出了挑战，到目前为止还没有充分解决的最先进的方法。 即使是最好的机器翻译系统翻译成阿拉伯语等语言，也经常产生不流利的翻译，缺乏适当的语法结构。 这个项目探索了新的方法，解决这些问题的统计MT框架，结合更深层次的建模语法和形态的发展。 虽然探索的方法在很大程度上是独立于语言的，本研究是在一个大规模的英语-阿拉伯语机器翻译系统的背景下进行的，该系统是利用LDC提供的大量语料库构建的。句法翻译资源，这些资源是从大量并行数据中自动获取的，并探索了几种替代途径，用于整合一个高精度的阿拉伯语形态分析和生成引擎提供的信息纳入MT框架。该项目还探讨了使用语法转换规则来改进阿拉伯语MT输出的语法的方法，这些语法转换规则模拟了英语和阿拉伯语之间的语法差异。我们的目标是开发一个英语到阿拉伯语的机器翻译系统，以机器翻译评估指标衡量，该系统产生的阿拉伯语输出比目前最好的系统要流利、语法和准确得多（如BLEU和METEOR），并由人工评估人员判断。从英语到阿拉伯语的高准确度全自动机器翻译的可用性对阿拉伯语人口具有很高的潜在价值，这种高质量的翻译成阿拉伯语的机器翻译还可能改善美国和国际公司在阿拉伯语世界的市场准入。

Machine Translation (MT) into morphologically-rich languages poses uniquechallenges that have so far not been adequately addressed in state-of-the-artapproaches. Even the best available MT systems into languages such as Arabicfrequently produce translations that are disfluent and lack proper grammaticalstructure. This project explores novel approaches that address these issues bythe development of a statistical MT framework that incorporates deeper levelsof modeling of syntax and morphology. While the methods explored are largelylanguage independent, the research is conducted and experimentally evaluatedwithin the context of a large-scale English-to-Arabic MT system constructedusing vast corpora available from LDC.The research in this project focuses on novel approaches for combiningsyntactic and non-syntactic translation resources that are automaticallyacquired from vast amounts of parallel data and on exploring severalalternative pathways for the integration of information provided by ahigh-accuracy morphological analysis and generation engine for Arabic into theMT framework. The project also explores methods for improving the syntax of MToutput in Arabic using syntactic transfer rules that model syntacticdivergences between English and Arabic. The goal is to develop anEnglish-to-Arabic MT system that produces significantly more fluent,grammatical and accurate Arabic output than the current best systems, asmeasured by MT evaluation metrics (such as BLEU and METEOR), and as judged byhuman evaluators.The availability of high-accuracy fully-automatic Machine Translation fromEnglish into Arabic has high potential value to the Arabic-speaking populationat large, by opening up access to all English content available over the web.Such high-quality MT into Arabic may potentially also improve access tomarkets in the Arabic-speaking world for US and international companies.