摘要
专有名词中的中国地名的自动识别是中文信息处理中要解决的一个重要问题 .完全依靠统计方法只能保证一定的召回率 ,而准确率偏低 .提出了一种基于变换的策略 ,可以有效地克服这一缺陷 .经测试 ,系统最终的准确率提高了 7% ,达到了 90 .9% .
The automatic recognition of Chinese place names, a special case of the recognition of Chinese special nouns, is an important task in Chinese information processing. The method based on statistical technique can only ensure the recall to some degree, but the precision is relatively low. An approach based on transformation is proposed, which can effectively overcome the deficiency caused by statistics. The performance of the approach is evaluated on a real data set, and the precision finally reaches 90.9%, improved by 7%.
出处
《软件学报》
EI
CSCD
北大核心
2001年第11期1608-1613,共6页
Journal of Software
基金
国家"九五"计划社科基金No.97@yy0 0 1- 2~~
关键词
中国地名
自动识别
专有名词
中文信息处理
计算机
Calculations
Evaluation
Information analysis
Mathematical transformations
Statistical methods