MTranslatability

Machine Translation of arbitrary input is difficult,but the output quality can be improved significantlyif writers create documents with MT in mind. Thisarticle deals with ``MTranslatability' – translatabilityof texts by MT systems. It identifies characteristics oftext that decrease MTranslatability and suggests waysto improve them. It also illustrates the effect of writingfor MTranslatability by showing before-and-after picturesof output from various commercially available MT systems,and gives an overview of tools that help identify and correctthe problems.     

基于纹理块与梯度特征的图像修复改进算法

针对目前图像修复算法在处理污损车牌图像时存在边缘修复不连续和效率不高问题,提出采用纹理块与梯度特征结合的改进的算法。新算法从两个方面进行改进：针对最优匹配块,加入梯度计算,优化匹配块的选择及预编辑,提高与环境的融合效果;采用了一种更新填充前端的修改方案,提高修复效率。测试结果表明,所提出的算法具有更好的计算效率,视觉感知方面具有更好的边缘连续性。     

Improvement of Translation Quality of English Newspaper Headlines by Automatic Pre-editing

Since the headlines of English news articles have a characteristicstyle, different from the styles which prevail in ordinary sentences, it is difficult for MT systems to generate high-quality translation for headlines.We try to solve this problem by adding to an existing system apre-editing module which rewrites headlines as ordinary expressions.Rewriting of headlines makes it possible to generate better translations which would not otherwise be generated, with little or no changes to the existing parts of the system. Focusing on the absence of a form of the verb be as a missing part of normal English, we have described rewriting rules for putting properly the verb be into headlines, based on information obtained by morpho-lexical and rough syntactic analysis.We have incorporated the proposed method into our English–JapaneseMT system, and carried out an experiment with 312 headlines as unknown data. Our method has satisfactorily marked 81.2% recall and 92.0% precision.