摘要
通过汉语自然语言与机器人进行人机交互无疑是一种高效便捷的导航办法。主要针对汉语路径自然语言的处理方法进行研究。首先搭建了10个非结构化的3D环境,针对所构建环境下机器人的导航任务完成了自然语言导航语料的收集,该语料库的来源以在校大学生为主体,辅以各年龄段不同职业的社会人士;然后采用NLPIR汉语分词系统对有效语料进行分词以及词性标注处理,最后为了提取用于导航的语义信息,定义了9种基本组块,并采用条件随机场（CRF）实现了语料的组块自动标注,实验结果表明该方法的组块标注准确率较高,为进一步提取导航语义打下了基础。
Man-robot interaction by Chinese natural language is an efficient and convenient method for robot navigation.In this paper,we mainly study the natural language processing method for Chinese route language.Firstly 10 unstructured 3D environments for robot navigation tasks were built,and then the natural language navigation corpus was collected based on these robot navigation tasks.The most of corpus collectors are college students,in addition to some people with different age and career.Secondly the NLPIR Chinese word segmentation system was used for segmentation and tagging.Finally in order to extract the semantic information for robot navigation,nine basic blocks were defined,and the conditional random field（CRF） model was employed to achieve the data block automatic annotation.The experimental results show that the accuracy of this chunking method is higher,which laid the foundation for further extraction of semantic navigation.
作者
王浩
景阳
王鲜惠
张超
潘蔚
陈奇
WANG Hao, JING Yang, WANG Xian - hui, ZAHNG Chao, PAN Wei, CHEN Qi （Electrical ＆ Electronic Engineering School,North China Electric Power University, Baoding 071000,China）
出处
《电脑知识与技术》
2016年第4期181-183,186,共4页
Computer Knowledge and Technology
基金
国家自然科学基金（No.61302163,No.61302105）
河北省自然科学基金（No.F2015502062）
关键词
自然语言处理
机器人导航
组块分析
条件随机场
natural language processing
robot navigation
chunking
conditional random field
