摘要
情态是一个复杂语义系统,其语义不确定性给其词义识别带来很大困难。语境与情态互动并对情态动词的词义有很大的限制作用,因此,揭示情态与语境互动限制关系对自然语言处理、机器翻译和智能语义识别特征优选研究都具有重要意义。本文基于形式概念分析理论和独有属性特征计算方法,通过数据挖掘发现情态动词词义与语境特征的互动关系,以及语境特征对情态动词词义的限制作用等知识。以英语情态动词must为目标词,针对其主要词义,从语料库中抽取150个样本例句作为对象,分别从must出现的语境中提取了32个特征,包括语义、句法、语用和话题特征作为属性,建立了表达must不同词义对象与其所具有的属性特征之间关系的形式背景。采用独有属性特征计算方法提取了must词义分类规则。通过对比分析其不同词义分类规则,发现了must与不同语境特征的互动限制关系。研究结果为自然语言处理和智能语义识别特征选择研究提供了有价值的依据。
English modal verbs are a complex semantic system and their indeterminacies in senses have brought great difficulties in the recognition of their word senses.Since contexts interact with and have a great restriction to the senses of modal verbs,it is an important issue to reveal the interactive and restrictive relations between modals and the contextual features for the studies of natural language processing,machine translation and feature selection and optimization in intelligent sense recognition.The interactive relations between the senses of modal verbs and the contextual features and the restriction of the contextual features upon the senses of modal verbs are investigated through data mining based on the theory of formal concept analysis and the approach of exclusive attributes feature calculation in this paper.The English modal verb “must” is taken as a target word,and in the light of its three main senses,150 sample sentences(50 for each sense) are extracted from a corpus as the objects,and 32 contextual features are extracted from the contexts as the attributes,including the semantic,syntactic,pragmatic and topic features.Then,the formal context representing the relations between the senses of “must”(objects) and the attribute features of “must” is established,and the related exclusive attributes are calculated and used as the rules for word sense classification.By comparing and analyzing the rules for different senses of “must”,the interactive and restrictive relations between “must” and the contextual features are discovered,which provides valuable references for the study of natural language processing and feature selection in intelligent sense recognition.
作者
于建平
付继林
白塔娜
李少辉
洪文学
YU Jianping;FU Jilin;BAI Tana;LI Shaohui;HONG Wenxue(School of Foreign Studies,Yanshan University,Qinhuangdao,Hebei 066004,China;Liren College,Yanshan University,Qinhuangdao,Hebei 066004,China;School of Electrical Engineering,Yanshan University,Qinhuangdao,Hebei 066004,China)
出处
《燕山大学学报》
CAS
北大核心
2019年第5期462-470,共9页
Journal of Yanshan University
基金
河北省社会科学基金资助项目(HB17YY047)
关键词
情态动词
语境特征
互动关系
独有属性特征
数据挖掘
modal verb
contextual features
interactive relation
exclusive attribute feature
data mining
