规则与统计相结合的兼类词处理机制  

Disambiguation Mechanism Using Rule Techniques and Statistics Techniques

作　　者：黄德根[1] 张丽静[1] 张艳丽[1] 杨元生[1] 

机构地区：[1]大连理工大学计算机系,辽宁大连116023

出　　处：《小型微型计算机系统》2003年第7期1252-1255,共4页Journal of Chinese Computer Systems

基　　金：国家自然科学基金 ( 60 14 3 0 0 2 )资助

摘　　要：兼类词处理是词性标注的关键所在 ,本文对兼类词排岐进行了研究 ,介绍了规则和统计相结合的排岐策略 .按照上述策略 ,实现了一个兼类词处理系统 .实验测试结果表明 ,利用规则与统计相结合的兼类词处理机制可以有效地提高排岐正确率和词性标注正确率 ,在封闭测试和开放测试中兼类词的排歧正确率分别达到了 93.91%和 91.16 % ,标注正确率分别达到了 97.85 %和 96 .71% .Syntactic category disambiguation is the key to part of speech tagging .In this paper, westudy the syntactic category disambiguation and introduce the disambiguation strategy using rule techniques and statistics techniques. With the above method, a system of disambiguation is materialized. The experimental results show the tagging accuracy is raised by using rule techniques and statistics techniques .The disambiguation accuracy of close test and open test is 93.91% and 91.16% respectively, and the overall accuracy is 97.85% and 96.71% respectively.

关 键 词：词性标注 兼类词 规则 n-元模型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]