基于子类的汉语动词SCF获取回退  

Backoff Techniques for Chinese SCF Acquisition Based on Sub-class

作　　者：韩习武[1] 赵铁军[1] 

机构地区：[1]哈尔滨工业大学计算机科学与技术学院,哈尔滨150001

出　　处：《计算机工程与应用》2005年第30期158-161,共4页Computer Engineering and Applications

基　　金：国家自然科学基金项目:1.汉语动词次范畴化自动获取技术的研究(编号:60373101);2.基于双语信息的英汉译文消歧技术研究(编号:60375019);国家863高技术研究发展计划(编号:2002AA117010-09)

摘　　要：研究证明基于动词子类的回退方法能够改进次范畴化(SCF)自动获取的性能,不同的子类划分方法起到重要的作用。本文利用《同义词词林》、《知网1.2版》和《现代汉语动词大词典》分别对30个测试动词进行分类,并以分类结果来改进汉语动词SCF自动获取的性能,实验结果证明:一、回退后的获取性能有了较为明显的提高;二、涉及句法成分较多的分类方法更适合汉语SCF自动获取;三、随机组合相关动词来构造回退子类是不合理的;四、上述分类都不是适合汉语动词SCF获取回退的最优方法。It has been proved that automatic subcategorization acquisition can be improved by backoff methods based on verb sub-classifications,and the concerned classification often makes great differences.This paper tries to partition 30 test verbs according to the lexical knowledge respectively in Semantic Thesaurns,Hownet 1.2 and Modern Chinese Verb Dictionary,and with the results backoffed the acquisition of Chinese verb subcategorization.The experiment shows that：a. the general performance has been promoted to certain extent;b.syntactic classification is more suitable for Chinese SCF acquisition;c.random classifications are irrational for backoff;and d.the above-mentioned classifications are not the best ones for Chinese SCF acquisition.

关 键 词：汉语动词 子类 SCF 回退 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]