Presidential Faculty Fellowship/Presidential Early Career Awards for Scientists and Engineers (PFF/PECASE)

该项目旨在演示使用基于词汇的参数化框架来适应自动词典构建并支持广泛的跨语言现象。这项工作不仅为解决大规模机器翻译工作背景下的跨语言差异提供了可行的解决方案，捕获了空间、时间和方面的知识，而且还减少了大规模机器翻译系统的开发时间。第一个任务包括确定自动字典支持MT的过程，并演示其对英语以外的语言（如阿拉伯语和韩语）的使用。第二个任务是扩展现有的机器翻译模型，通过容纳空间关系的组合衍生的中间语言来包含更广泛的不同语言现象。最后一个任务是扩展当前的词汇语义表示，以包括时间和方面的知识。

The project is oriented to demonstrate the use of a lexical-based parametric framework to accommodate automatic dictionary construction and to support a broad range of cross-linguistic phenomena. This effort not only provides a viable solution to resolving cross-linguistic distinctions in the context of a large-scale MT effort, capturing spatial, temporal, and aspectual knowledge, but also reduces the time for the development of large scale machine translation (MT) systems. The first task consists of determining the procedures for the automatic dictionary support for MT, with demonstration of its use to languages other than English such as Arabic and Korean. A second task consists of the extension of an existing MT model to include a broader range of different linguistic phenomena through an interlingua which accommodates the compositional derivation of spatial relations. A final task is the extension of current lexical-semantic representation to include temporal and aspectual knowledge.