摘要
基于定长序列的双向LSTM标注模型,提出强化中心字模型和强化逆向序列模型的分词优化方法.在Bakeoff2005微软亚洲研究院语料上的实验表明,优化方法的分词效果得到了提升.提出的优化模型,对于其他序列标注任务,具有一定的借鉴作用.
The method of center character enhancement and the method of reverse sequence enhancement were proposed based on fixed length sequence BLSTM.The experimental results on Bakeoff 2005 Microsoft Asia Research Institute showed that,the segmentation effect of the optimization method was improved.The proposed optimization model could be used as a reference for other tasks.
作者
郭正斌
张仰森
GUO Zhengbin;ZHANG Yangsen(Institute of Intelligent Information Processing,Beijing Information Science and Technology University,Beijing 100192,China)
出处
《郑州大学学报（理学版）》
CAS
北大核心
2018年第2期97-101,共5页
Journal of Zhengzhou University:Natural Science Edition
基金
国家自然科学基金项目(61370139
61602044)
北京市教委科研计划一般项目(KM201711232022)
