A Study On Model Of Recognizing Illocutionary Point

The model of recognizing illocutionary point (MRIP), respectively based on indirect speech acts theory as ontology, the theory of conversational implicature as conversational level, and the theory of computational linguistics and artificial intelligence, is used to construct the platform of natural language understanding(NLU) and computational pragmatics. Its research aim is to discover the detailed mechanism from literal illocutionary act to primary illocutionary acts and realize NLU on pragmatics layer.BDI model is the computational theory of Searle's indirect speech acts theory, and it has developed as a leading model in agent area. In this dissertation, IBDI is proposed as a model of recognizing illocutionary point. On one hand, it is expected to amend the two deficiencies in BDI model, i.e., simple classification of speech acts and no generalization of illocutionary acts and illocutionary points. The new model will make ontological classification and conversational classification for illocutionary points, and increase the sorts of illocutionary points, give the formal representation for illocutionary point, increase some of the plan inference rules, and present a solution for recognizing direct and indirect illocutionary point by two types of partial plan. On the other hand, IBDI model inherits the inference tradition of BDI model on paradigm. Based on the plan inference and construction rule, the heuristic rule and control system, it is enabled to recognize illocutionary points and illocutionary acts by prior partial plans and obstacle detection.In this thesis, we also give HMM model of recognizing illocutionary point, as a new Stat. solution about pragmatic explanation, its construction path is, depending on the sentence lists, including word lists and rhythm, and HMM:λ=(A,B,π), and arithmetic viterbi, to recognize internal prior illocutionary point lists: IP=(ip1,ip2,,…,ipT). Therein, to compute the emission probability of word lists by word N-gram model, to compute the emission probability of rhythm by CART method, and to compute the transfer probability of internal state by illocutionary points FSA based on point N-gram.Based on analysis to context theory and philosophy of language, this dissertation proposes the PAP solution to illocutionary point and the FOC solution to context, and takes this solution as the philosophical base of MRIP model. Moreover, based on the analysis of character of illocutionary points including public, accessibility, privacy, the dissertation presents illocutionary logic, default logic ,and arithmetic of probability as logical and mathematical base of MRIP model. Based on philosophy, logic and mathematic foundation, this dissertation presents the innovative theory frame of MRIP model, and on this frame foundation, also presents two theory paradigms: logical inference paradigm and probability paradigm.