1.1 METEOR简介 METEOR(Metric for Evaluation of Translation with Explicit ORdering)是一种用于评估机器翻译质量的指标,由Banerjee和Lavie在2005年提出。与BLEU不同,METEOR不仅考虑词汇的精确匹配,还引入了同义词、词干和词序等因素,从而提供更全面的评估。 1.2 METEOR的计算方法 METEOR的计算基于以下几个步骤: 精确匹配...
2004年,卡内基梅隆大学的Lavir提出评价指标中召回率的意义,基于此研究,Banerjee和Lavie(Banerjee and Lavie, 2005)发明了基于单精度的加权调和平均数和单字召回率的METEOR度量方法,目的是解决BLEU标准中的一些固有缺陷。简单说,该指标考虑了基于整个语料库上的准确率和召回率,而最终得出测度。 METEOR (Metric for Evalua...
METEOR,全称为Metric for Evaluation of Translation with Explicit ORdering,中文意思是具有明确排序的翻译评估指标。METEOR的定义和概念主要基于单词级别的准确率和召回率,以及对词序的惩罚,来计算候选文本和参考文本之间的相似度。 METEOR的特点是它不仅考虑了单词的精确匹配,还考虑了词干、同义词和其他语言变体的匹配。...
METEOR: Metric for Evaluation of Translation with Explicit ordered 用于机器翻译的准确性评估,预测集 predictions 和参考集 references 可以是一对一的,也可以是一对多的。 METEOR 使用 单元词组精确度、单元词组召回率、碎片惩罚 三部分的组合来计算分数。 引入了较为灵活的对齐机制。 用通俗的话解释,METEOR 在评估...
We discuss the initial design of the METEOR metric, subsequent improvements, and performance in several independent evaluations in recent years.doi:10.1007/s10590-009-9059-4Alon LavieMichael J. DenkowskiSpringerMachine TranslationA. Lavie and M.J. Denkowski, “,The METEOR Metric for Automatic ...
We describe METEOR, an automaticmetric for machine translation evaluationthat is based on a generalized concept ofunigram matching between the machineproducedtranslation and human-producedreference translations. Unigrams can bematched based on their surface forms,stemmed forms, and meanings; furthermore,...
(2) ''Pho- tometry'' produces calibrated meteor light curves with associated astrometric data and (3) ''VideoCal'' provides instrument specific saturation compensa- tion. These components are supplemented with numerous utilities for view- ing, plotting, sorting, browsing, and quality controlling ...
These are simply random bits of solar system debris intercepted by our planet, which are part of the several metric tons of space dust that NASA estimates is swept up by Earth every day. At certain times of the year, ...
On any particular night of the year, it's possible to see a few random meteors streak by overhead. Scientists refer to these assporadics. These are simply random bits of solar system debris intercepted by our planet, which are part of the several metric tons of spac...
intuitive requirement is that the metric have very high correlation with quantified human notions of MT quality. Furthermore, a good metric should be as sensitive as possible to differences in MT quality between different systems, and between different versions of the same system. The metric should...