CORC  > 北京大学  > 信息科学技术学院
Improved parsing with taxonomy of conjunctions
Li, Dongchen ; Zhang, Xiantao ; Wu, Xihong
2014
英文摘要Incorporating knowledge for training a parser has been shown to remedy the weaknesses of probabilistic context-free grammar. Previous parsing systems have exploited content words semantic resource and word-formation knowledge. However, they are limited in that they do not take into account conjunction category refinement, which stands out to be helpful in predicting the syntactic structure and syntactic label in Chinese. We define a conjunction taxonomy representing intrinsic syntactic constraints, and show that refined categories in the taxonomy for conjunctions contribute to improved parsing performance. The taxonomy is used to supervise the splitting of these refined tags, and the automatic hierarchical state-split approach is employ to compensate the limitation in the scope and refinement degree of the taxonomy. The experiments are carried out on Penn Chinese Treebank, which show that our method can improve parsing performance significantly. ? 2014 IEEE.; EI; 0
语种英语
DOI标识10.1109/ChinaSIP.2014.6889199
内容类型其他
源URL[http://ir.pku.edu.cn/handle/20.500.11897/329982]  
专题信息科学技术学院
推荐引用方式
GB/T 7714
Li, Dongchen,Zhang, Xiantao,Wu, Xihong. Improved parsing with taxonomy of conjunctions. 2014-01-01.
个性服务
查看访问统计
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。


©版权所有 ©2017 CSpace - Powered by CSpace