基于CRF模型的英文词性标注研究  

Research on English part of speech tagging based on CRF model

作　　者：刘星宇 宁慧 张汝波[3] LIU Xingyu;NING Hui;ZHANG Rubo(College of Software,Harbin Engineering University,Harbin 150001,China;College of Computer Science and Technology,Harbin Engineering University,Harbin 150001,China;College of Mechanical and Electronic Engineering,Dalian Minzu University,Dalian 116600,China)

机构地区：[1]哈尔滨工程大学软件学院,黑龙江哈尔滨150001 [2]哈尔滨工程大学计算机科学与技术学院,黑龙江哈尔滨150001 [3]大连民族大学机电工程学院,辽宁大连116600

出　　处：《应用科技》2021年第1期25-30,35,共7页Applied Science and Technology

基　　金：国家自然科学基金项目(61673084)。

摘　　要：针对如何使用适当的模型或结构使得词性标注结果准确率提升的问题,对隐马尔可夫模型和条件随机场模型进行了深入研究和实验,使用条件随机场的不同特征方程进行了多组实验,并对比了每组实验的准确率。实验结果表明,条件随机场对于解决英文词性标注问题有着更大的优势;将共性的特征与相对具体的后缀特征结合使用所达到的词性标注准确率最高。In order to improve the accuracy rate of part of speech tagging results by using appropriate model or structure,the researches and experiments of hidden Markov model and conditional random field model are done in depth in this paper.Meanwhile,multigroup experiments are carried out,using different characteristic equations of conditional random field.The accuracy rates of these group experiments are compared.The results show that the conditional random field has a greater advantage in solving the problem of English part of speech tagging.The accuracy rate of part of speech tagging reaches the highest when the common features and specific suffix features are used in combination.

关 键 词：条件随机场 隐马尔可夫模型 词性特征 维特比算法 词性标注 自然语言处理 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]