Analysis For Deep Semantic Structures Of English Modal Verb Can

Word sense disambiguation(WSD) is a critical subject in the field of natural language processing. It has the great significance for information retrieval, machine translation, text categorization, speech recognition, etc. There are many relevant researches on word sense disambiguation at home and abroad, including nouns, verbs, and prepositions, etc.However, few of researches on word sense disambiguation(WSD) is focused on the English model verbs. The English modal verb can is the research object of the thesis. The modal verbs are applied to express human being’s mood, attitude, and feeling. Semantic indeterminacy and subjectivity are obvious. The research of the modal verbs is always the difficult topic in the field of WSD.Formal Concept Analysis is the theoretical foundation, and corpus-based approach and the generation tool of partial ordered structure are utilized in the thesis. Based on the WSD model, knowledge discovery of can on the deep semantic structures and interior structures is illustrated. Deep semantic structures refer to the structure of implicit linguistic and contextual features when a word has a certain meaning. And those are obtained by analyzing the structure and context of sentences. A natural language corpus with 800,000 words is established. And can is categorized into four categories according to authoritative dictionaries. Then, one training set and one testing set are generated, and the rules are extracted from the diagram to calculate the accuracy of WSD of can. The accuracy of WSD reaches 96.5%. These research results demonstrate the method and process of the study are scientific and feasible. Five kinds of syntactic features are introduced as the deep semantic features. The deep semantic structures are studied by observing the combination and collocation among different syntactic features, and the relevance and independence between semantic features and syntactic features. The deep semantic structures of can are analyzed from three perspectives: the structural partial ordered attribute diagram(SPOAD), the structural partial ordered object diagram(SPOOD), and the formal context. The attributes behind the different complex senses of can is studied to explore comprehensively the deep semantic structures of can.The major findings are discovered as follows: 1) the different characteristics of each category are obtained by the extracted rules. And the essential attributes and language characteristics are found to distinguish one category from the other three categories. 2) the semantic features play an important role in WSD in general. It takes on the gradient distribution from semantic features to syntactic features. The semantic features of can have universality, and the syntactic features of can have specificity. 3) the collocation affinity and tendency of different attributes in each category of can are discovered.The systematic and comprehensive semantic structural analysis on English model verb can from deeper perspective is made in the study. The research of the English modal verbs is extended from superficial to underlying, and the semantic study of the model verbs is broadened. The study provides the theoretical and practical foundations for the deep study of other complex or vague words.