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独特的挑战，迄今为止还没有在最先进的方法充分解决。即使是最好的机器翻译系统，如阿拉伯语，也经常产生不连贯的翻译，缺乏适当的语法结构。该项目通过开发一个统计机器翻译框架来探索解决这些问题的新方法，该框架结合了更深层次的语法和形态学建模。虽然所探索的方法在很大程度上是独立于语言的，但研究是在使用LDC提供的大量语料库构建的大规模英语到阿拉伯语的机器翻译系统的背景下进行和实验评估的。本项目的研究重点是结合从大量并行数据中自动获取的句法和非句法翻译资源的新方法，并探索将高精度阿拉伯语形态分析和生成引擎提供的信息整合到theMT框架中的几种替代途径。该项目还探索了使用语法迁移规则改进阿拉伯语MToutput语法的方法，该规则模拟了英语和阿拉伯语之间的句法差异。目标是开发一个英语到阿拉伯语的机器翻译系统，根据机器翻译评估指标（如BLEU和METEOR）和人类评估人员的判断，该系统比目前最好的系统产生更流畅、语法和准确的阿拉伯语输出。从英语到阿拉伯语的高精度全自动机器翻译的可用性对广大讲阿拉伯语的人口具有很高的潜在价值，因为它开放了对网络上所有可用英语内容的访问。这种高质量的阿拉伯语翻译也可能潜在地改善美国和国际公司进入阿拉伯语世界的市场。

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.