本文的内容是在1998年中文信息处理国际会议小组讨论上发表的基础上加工而成。本文主要讨论了我公司最近研发的英日自动翻译系统中日语生成模块的一些理论和实践问题。主要包括以下几个内容: 1. 语法过滤机制:是该生成模块的开发的核心思想; 2. 格助词的处理问题,这是日文生成成败的关键问题; 3. 核心句转换:整个生成过程就是一个从非核心句到核心句的转换过程; 4. 助动词的处理,P-树结构向D-树结构转换的重要组成部分; 5. 否定结构的处理,是该生成模块最困难的问题之一。

This technical report , developed from the short presentation at‘1998 ICCIP , is concerned with the phases of Japanese generating module in our English-Japanese translation system. Its fundamental focus is on the following points after giving an overview of the system: 1. Grammatical filter mechanism , the fundamental idea when developing our generating module. 2. Postposition handling mechanism , the central issue when attempting to generate grammatical Japanese sentences. 3. Kernel sentences transformation , the way of transforming original trees into object trees. 4. Auxiliary handling mechanism , one of vitally important components when completing the transformation from P-tree to D-tree. 5. Negation handling mechanism , the toughest phenomenon when handling.