基于规则和混合统计模型的中文地名识别研究  

Recognition of Chinese Place Names Based on Rules and Mixed Statistical Models

作　　者：林雄鹰 黄宇光[1] 

机构地区：[1]江南计算技术研究所,无锡214000

出　　处：《高性能计算技术》2012年第1期31-35,共5页

摘　　要：为了准确识别中文地名，本文提出了一种基于妞则和混合统计模型的中文地名自动识别方法。该方法的特点在于：融合了两种统计模型的优势进行地名识别，其中隐马尔可夫从整体上对中文地名的识别进行约束，最大熵有效融合了地名的多种约束信息，并利用规则对识别结果进行补充和修正。实验结果表明，本文提出的方法是有效的，开放测试的召回率、准确率和F值分别为89．67％、85．45％和87．51％。Aiming at solving recognition of Chinese place names, this paper proposed a method of automatic recognition of Chinese place names, which combines mixed statistical models with rules. The characteristic of this method is that it combines with two statistical models--the Hidden Markov Model, which restrict the recognition of Chinese place names as a whole, and the Maximum Entropy, which can syncretizes multiple restriction, and uses rules to recall and correct the recognition result. The experiment result shows that the algorithm is efficient. In open test, the recal, precision and F-measure achieves 89.67%, 85.45% and 87.51%.

关 键 词：中文地名识别 隐马尔可夫 最大熵 规则 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]