平仄信息对中文词法分析的影响

来源 :中国中文信息学会 | 被引量 : 0次 | 上传用户:lsssml1990
下载到本地 , 更方便阅读
声明 : 本文档内容版权归属内容提供方 , 如果您对本文有版权争议 , 可与客服联系进行内容授权或下架
论文部分内容阅读
  词法分析,作为自然语言处理领域的基础性研究课题之一,其效果直接影响自然语言处理后续的工作。本文从汉语语音的特色出发,利用机器学习的方法,学习汉语句子中词语的平仄信息,研究平仄信息对词法分析的影响。并分别在人民日报语料和宾州中文树库语料上设计实验,实验结果证明了平仄信息特征确实能够大幅度提高中文词法分析的精度。
其他文献
@@ This paper, under the minimalist framework and by virtue of the assumption of FP (or pP), (cf. Van Riemsdijk1990, Rooryck1996, Ayano2000,Thomas2004, Gehrke2007, among others), tries to provide answ
会议
@@ Current EAP courses were criticised for their overemphasis on the general academic core rather than the disciplinary genericfeature (Bhatia; 2004), in spite of using subject content as a vehicle of
会议
@@ The middle construction, like passives, can be impersonal and have its external argument unexpressed even when the verb is of transitive form. However, it differs from passives in that the verb doe
会议
@@ The legacy of Handbook of Biolinguistics (Meader & Muyskens 1950/1962) in contemporary biolinguistics research can be partially highlighted from four aspects. First in the history of biolinguistics
会议
@@ Is rule-based processing and acquisition psycho-linguistically real for EFL learners? To explore this, the paper reports two psycholinguistic experiments investigating the processing mechanism of r
会议
According to the Copy Theory of Movement(Chomsky 1995), a lexical item and its copy can’t be phonetically realized at the same time, otherwise the derivation will crash at PF because the lexical items
会议
@@ This paper presents evidence to support the idea that the syntax-pragmatic interface approach (Sorace 2007, Yoshimura et al. 2011) can provide a plausible account for the L2 acquisition of locality
会议
@@ Grodzinsky and Reinhart (1993) and Reinhart (2006) claim that the construction and comparison of a set of competing derivations (known as reference-set computation) comes with a processing cost, an
会议
本文介绍了都柏林城市大学(DCU)在第七届全国机器翻译研讨会(CWMT’2011)上的参赛系统。该系统由基本短语翻译系统、层次短语翻译系统、系统融合三部分组成。本系统在日中新闻子任务中取得了较好的成绩。
本文详细描述了中国科学院合肥物质科学研究院智能机械研究所参加的第七届全国机器翻译研讨会(CWMT2011)评测中的第7个项目--维汉新闻领域的翻译评测系统。在评测中我们共使用了四个翻译系统:基本系统是一个标准的基于短语的统计机器翻译系统;第一个对比系统是融入了源语言形态信息的短语翻译系统,利用递推式的最小描述长度原理对源语言进行非监督式的形态切分,并结合逆向最大匹配词缀的方法获取形态信息;第二个对