Analysis Of The Contextual Features Influencing The Semantic Mergers Of English Modal Verb May

Semantic indeterminacy as a common phenomenon in natural language is also a key and difficult problem in natural language processing(NLP). In recent years, the technology of artificial semantic disambiguation has witnessed great development. However, the current research scope is concentrating on notional words with relatively clarified meanings. Researches on modal verbs are also mainly about superficial semantic meanings. However, English modal verbs features semantic fuzziness and contextual sensitivity, which thus makes the exploration of implied contextual features with interdisciplinary methods significantly important to artificial semantic disambiguation.Semantic indeterminacy includes gradience, ambiguity and merger. Semantic merger is a phenomenon of semantic convergence of two meanings of a word in a single context. Based on Formal Concept Analysis theory and adopting structural partial-ordered attribute diagram(SPOAD) mechanism, this thesis is aimed to analyze the implied contextual features that have an influence on the semantic mergers of English modal verb may and thus realize the purpose of semantic disambiguation. This thesis is based on a corpus of 1.8 million words adopting Wconcord to get twelve semantic features. Besides, ten syntactic features that may have an influence on the semantic meanings are also extracted. Through the vectorization of the extracted linguistic features and with the help of SPOAD, a disambiguation model for English modal verb may is obtained. The self-check accuracy rate by leave-one-out validation of this disambiguation model reaches 96.25%, and the accuracy rate of five-fold cross validation is 89.38%±3.57%, demonstrating the validity of this model.Based on the valid WSD model, rules are extracted from the diagram for English model verb may. Through the analysis of extracted rules and the attribute features, it can be found that the semantic merger of English model verb may occurs mainly in two situations: 1) the MIs of may with its subject and main verb are all high value; 2) the topic of the sample sentence indicates the potential result of an event or a proposition and at the same time implies that it will happen in the future. As a result, the subject and main verb related with may, “topic implying results”, and “topic with implied future” are influential contextual features for semantic merger of may.Based on the Formal Concept Analysis theory and employing SPOAD method, this thesis analyzes the contextual features that contribute to semantic mergers of English modal verb may, providing a new aspect for the research of semantic ambiguity from the perspective of implied semantics. The construction of disambiguation model for semantic merger of English modal verb may not only effectively identifies semantic mergers of may in different context, realizing automatic tagging of semantic meanings, but also analyzes the implied reasons that lead to the semantic mergers and thus make contribution to natural language processing and linguistic research.