论文部分内容阅读
多语言信息存取(MLIA:Multilingual Information Access)是许多数字图书馆希望提供的一种服务。本文从MLIA的概念入手,探讨了机器翻译(MT)在实现数字图书馆的多语言信息存取服务中的作用。为了评测目前MT系统的性能以确定它们能否应用于数字图书馆的多语言信息存取,我们开发了一个多语种的元数据机器翻译综合评价平台HeMT(Human Evaluation of Machine Translation.http://txcdk-v10.unt.edu/HeMT/)。利用HEMT平台6大功能模块,评价了3个在线机器翻译系统所译的2000条元数据记录。这些记录来自北德克萨斯大学(UNT)图书馆和德克萨斯州历史门户(The Portal to Texas History),在报告评价结果的基础上,对所发现的机器翻译的错误进行了分析,文章最后提出了使用多引擎机器翻译(Multi-engine Machine Translation)技术以提高元数据记录翻译性能的设想。
Multilingual Information Access (MLIA) is a service many digital libraries want to provide. This article begins with the concept of MLIA and explores the role of machine translation (MT) in realizing the multilingual information access service of digital library. In order to evaluate the performance of current MT systems to determine whether they can be applied to multilingual information access in digital libraries, we developed a multilingual Meta-Machine Meta-Machine Translation Evaluation Platform HeMT (Human Evaluation of Machine Translation.http: // txcdk-v10.unt.edu/HeMT/). Using the 6 functional modules of HEMT platform, 2,000 metadata records translated by 3 online machine translation systems were evaluated. These records, from the University of North Texas (UNT) Library and The Portal to Texas History, analyze the findings of machine translation errors, Finally, the article proposes the idea of using Multi-engine Machine Translation technology to improve the translation performance of metadata records.