Quantification Research On The Fuzzy Semantics Based On MMTD

Natural language understanding, whose research is greatly influenced by the semantic fuzziness, is a core subject in the domain of Artificial Intelligence. While quantification research on the fuzzy semantics is always a hot issue of computer science society and linguistic field. This thesis seeks to provide a new way of dealing with fuzzy semantics, which is based on MMTD (measuring of medium truth degree).In the beginning of this thesis, the research of Natural language understanding and the study status of quantification of the semantic fuzziness are briefly introduced. After that, the content of MMTD is separately described in one-dimensional and multi-dimensional conditions.On this basis, the quantified objects, the quantifying processes, the set of syntactic and semantic rules of the MMTD semantic fuzziness quantifying algorithm are described in great detail. The truth degree function of Chinese adverbs of frequency is set up according to the result of numerical value of English adverbs of frequency. And the truth degree function of Chinese adverbs of degree is also established after setting their values. The concept of adverbs connection is put forward, after bringing in the adjective appraise classification and the emotional coefficient, quantification method of adverbs connection is discussed. The measuring of truth degree of complex sentences is carried out at the end of this part.Furthermore, this paper establishes quantification frame of fuzzy semantic based on MMTD, designs the whole flow and some programs and perform related experiments. The results indicate that quantification of fuzzy semantics of Chinese based on MMTD is effective.The work in this thesis is an attempt of quantification of fuzzy semantics of Chinese based on logic truth degree, which tries to find a new way in studying natural language understanding of artificial intelligence.