基于动态规划算法的专有名词切分  

Segmentation of Named Entity Based on Dynamic Programming

作　　者：李成城[1] 赵述芳 刘建毅[1] 钟义信[1] 

机构地区：[1]北京邮电大学智能科学技术中心 [2]中国人民解放军军械工程学院军械技术研究所,河北石家庄050000

出　　处：《计算机应用研究》2005年第7期78-80,共3页Application Research of Computers

摘　　要：提出了一种对专有名词字串进行自动切分的方法,该方法对大量通用形式的专有名词进行统计分析,根据专有名词用字的出现位置和出现次数来分析,并把出现概率转换为估价值。则根据中文专有名词的特点,将中文专有名词的切分问题转换为决策树求最优解问题。采用动态规划法选取最优路径,从而确保结果的整体估价值最优。以中文姓名为例,实验表明该方法具有很快的速度和很高的正确率,用10个名字的字符串进行测试正确率为99.66%以上。This paper put s forward an approach to segment Chinese name when is put together. It carries out a statistical analysis of a great number of most commonly used named entity a nd sets up a corpus of name according to the position and the frequency of the occurrences of the characters used in named entity, and also transforms the frequency of the occurrences into evaluation value. Based on the Chinese characteristics used in Chinese name, it changes the segmentation of Chinese name into decision process of optimality. Dynamic programming is used to choose the optimum method. AS an example, Chinese names are used in our experiments to show that this approach works at a fairly high speed and with a high precision. The precision amounts to over 99.66% when 10 names written together are tested. Moreover, the approach can be further used in Chinese words segmentation.

关 键 词：人工智能 自然语言处理 专有名词识别 动态规划 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]