【摘要】 英语从句识别是句法分析的基本问题,它是进一步进行英汉机器翻译的基础。提出一种基于最大熵原理的英语从句识别方法,即先将提取后的特征表示为最大熵模型形式并对其编码,再采用最大熵原理完成最后的识别过程。采用该方法进行完整从句识别的正确率和召回率分别达到80.92%和62.36%,实验表明,该方法的正确识别率远高于CoNLL01的底线标准,是一种简单有效的从句识别方法。更多还原

【Abstract】 As the basis of the syntax analysis,clauses recognition is an important step to the English-Chinese machine translation.A method based on the maximum entropy model for clauses recognition is proposed in this paper.First,the extracted features are described in maximum entropy style and are encoded.The final recognition is carried out according to the maximum entropy principle.Accuracy rate of a complete clause reaches 80.92% with a high recall of 62.36%.The great improvements of the baseline of CoNLL01 show the simplicity and effectiveness of the proposed method.更多还原