基于转移的音字转换纠错规则获取技术  

AUTOMATIC RULE ACQUISITION BASED ON TRANSFORMATION FOR A CHINESE SYLLABLE TO CHARACTER CONVERSION SYSTEM

作　　者：关毅[1] 王晓龙[1] 张凯[1] 

机构地区：[1]哈尔滨工业大学计算机科学与工程系

出　　处：《计算机研究与发展》1999年第3期268-273,共6页Journal of Computer Research and Development

基　　金：国家"八六三"高技术计划基金

摘　　要：文中描述了一种在音字转换系统中从规模不限的在线文本中自动获取纠错规则的机器学习技术．该技术从音字转换结果中自动获取误转换结果及其相应的上下文信息，从而生成转移规则集．该转移规则集应用于音字转换的后处理模块，使音字转换系统的转换正确率进一步提高，并使系统具备了很强的灵活性和可扩展性．Here described is a new approach to automatically learning error correct rules from online text of unlimited size for a syllable to character conversion system.It is shown that this method not only can improve the precision of the result of syllable to character conversion, but also can enhance the flexibility and expansibility of the system.This automatically learning model together with the former syllable to character conversion system forms a new type of rule statistic hybrid syllable to character conversion system.

关 键 词：音字转换 纠错规则 机器学习 汉字信息处理 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]