Font Size: a A A

Study On Feature Of Attributes Of English Modal Verb Must

Posted on:2018-01-26Degree:MasterType:Thesis
Country:ChinaCandidate:M F MiaoFull Text:PDF
GTID:2405330566960224Subject:Foreign Linguistics and Applied Linguistics
Abstract/Summary:PDF Full Text Request
English modal verbs have a very complex semantic system acting as the verbs to convey the feelings of human beings.The senses of modal verb are influenced by semantic,syntactic and other various kinds of factors,which makes modal verbs more difficult to recognize and describe.Therefore,researches on modal verb have been a hot and tough issue in the natural language processing.Attribute features are features of attributes in formal context.Features of attributes can describe the relations between objects and attributes more scientific.And the complicated relations between objects and attributes can be described by features of attributes.This study provides a new perspective to distinguish and understand the different senses of modal verbs.This thesis is based on a corpus with 3.5 million words.The extraction of features of attributes is conducted based on formal context and attribute partial ordered structure diagram.English modal verb must is chosen as the target word.The senses of must are divided into three kinds including root obligation,root necessity and epistemic inference.The mutual information between must and adjacent subjects and verbs are treated as semantic features.12 syntactic features relating to time,aspect,voice and context are extracted.150 sample sentences in which must occurs are selected as objects,and 12 semantic features and 12 syntactic features are selected as attributes for constructing the formal context of must.Then the corresponding attribute partial ordered structure diagram for the word sense disambiguation of must is generated.After ten-fold cross validation,the accuracy of attribute partial ordered structure diagram is 95.07%±0.91%.Based on the model of word sense disambiguation of must,the features of attributes such as simple class exclusive attribute,composite class exclusive attributes,maximum common attribute,and opposite attribute are extracted and analyzed.Authority is the maximum common attribute and simple class exclusive attribute of must(obligation).Passive voice is the simple class exclusive attribute of must(necessity).Perfect aspect,progressive aspect and existential subject are the simple class exclusive attribute of must(inference).The maximum common attribute and composite class exclusive attributes can be extracted form attribute partial ordered structure diagram.The opposite attribute is extracted from the formal context.This thesis researches on the features of attributes based on the formal context and attribute partial ordered structure diagram.The results of this study can help optimize the selection of features for word sense disambiguation of must.These findings also provide significant evidence for both linguistic study and natural language processing.
Keywords/Search Tags:modal verbs, attribute partial ordered structure diagram, formal context, features of attributes
PDF Full Text Request
Related items