摘要
针对军用要图自动标绘对地名识别高精度的要求,紧扣"作战文书行文规范"这一特点,将句类分析之标点准则(语句的首尾边界是首块和末块的天然首尾边界)、介词准则(汉语的语块移位标志符由介词充当)、动词准则(充当特征块的动词是其它语块的天然边界)等应用于作战文书地名识别,采用基于规则的方法加以实现,实验结果F值在88%-97%之间。因此,作为作战文书向军用要图自动转换的预处理,地名识别这一结果是可用的。
To meet the need of highly-accurate recognition of place names oriented to automatic military charting,and based on the Para-standardizing characteristic of operation documents,some SCA-oriented rules,are applied into the recognition of place names in operation documents for the first time,i.e.punctuation rule（the start of a sentence is the start of its first chunk,and the end is its last chunk）,preposition rule（the identifiers of chunks shift are preposition）,verb rule（the verbs as Eigen chunk are naturally the boundaries of other chunks）,etc,and the result of F-value is between 88 and 97 percent.As one of the preprocessings of auto-transferring to military chartings from operation documents,the rule-based way is feasible and the result is available.
出处
《计算机工程与设计》
CSCD
北大核心
2013年第8期2903-2907,共5页
Computer Engineering and Design
关键词
中文地名识别
语块边界感知
句类分析
作战文书
概念层次网络理论
chinese place name recognition
perception of chunks＇ boundaries
SCA（sentence category analysis）
operation document
HNC theory
