基于联合模型的中文嵌套命名实体识别  

Chinese Nested Named Entity Recognition Using a Joint Model

作　　者：尹迪[1] 周俊生[1] 曲维光[1] 

机构地区：[1]南京师范大学计算机科学与技术学院,江苏南京210023

出　　处：《南京师大学报（自然科学版）》2014年第3期29-35,共7页Journal of Nanjing Normal University(Natural Science Edition)

基　　金：国家自然科学基金(61272221;61472191);江苏省社科基金(12YYA002)

摘　　要：中文嵌套命名实体识别是自然语言处理中一个比较困难的问题.针对传统的序列化标注方法的不足,本文提出了一种新的基于联合模型的中文嵌套命名实体识别方法,该方法将嵌套命名实体识别看作是一种联合切分和标注任务.联合模型用一种改进的beam search算法作为系统的解码算法,并采用一种在线学习算法平均感知器算法作为训练算法,获得了较快的收敛速度和较好的识别效果.实验结果表明基于联合模型的方法对嵌套命名实体识别取得了更好的效果.Chinese nested named entity recognition is a very difficult problem in natural language processing. This paper presents a novel method based on a joint model,which treats the recognition of Chinese nested named entity as a task of joint word segmentation and labeling. The proposed method exploits an improved beam search algorithm as decoding algorithm,and uses the averaged perceptron algorithm as training algorithm,attaining fast convergence during training. The experimental results show that the joint model achieves better performance than two baseline systems using the traditional sequence labeling models.

关 键 词：嵌套命名实体识别 序列化标注模型 联合模型 感知器算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]