Research On Semantic Disambiguation Of Human-Machine Dialogue Based On Subject Dictionary

Natural language understanding is the first step to realize human-machine dialogue.In the human-machine dialogue system,mature word segmentation dictionary ensures the correct segmentation of natural language and avoids the problem of semantic ambiguity caused by word segmentation ambiguity.However.,there are still two problems in word segmentation dictionary in human-machine dialogue system at present:1.the structure of word segmentation dictionary and the generalization of dictionary content;2.there are loopholes in the construction of word segmentation dictionary and lack of corresponding measures to ensure the accuracy of word segmentation.Therefore,this paper establishes a double-word hash subject dictionary to solve the problems of generalization of dictionary structure and dictionary content;secondly,an optimization algorithm of word segmentation dictionary is proposed to solve the problem of imperfect construction of word segmentation dictionary in human-machine dialogue system.Finally,the previous work is applied to the human-machine dialogue system to maximize thesemantic ambiguity caused by word segmentation ambiguity.The main work ofthis paper is also the innovation of this paper as follows:1.The construction of a double word hash subject dictionary solves the problems oflack of universal dictionary mechanism and generalization of content in human-machine dialog system.By comparing the spatial capacity and search speed of word segmentation dictionaries with different mechanisms,the double word hash mechanism is finally selected to construct a subject dictionary.The subject dictionary is constructed by using dictionary resources and corpus training mode which solves the problem of semantic ambiguity in human-machine conversation caused by the limitation of universal dictionary coverage and the lack of dictionary mechanism.2.The optimization algorithm of word segmentation dictionary is proposed to solve the problem of imperfect construction of word segmentation dictionary in subject-oriented human-machine dialogue system.In order to eliminate the useless words and make up for the loopholes of word segmentation,the maximum inverse matching algorithm is used to solve the problem of word segmentation ambiguity in reverse matching between the results of word segmentation system and the subject dictionary based on double-word hash mechanism.Finally,the word segmentation ambiguity resolution algorithm is used to deal with the problem of word segmentation ambiguity in reverse matching.3.Subject dictionary and dictionary optimization algorithm are applied to the human-machine conversation system to maximize the semantic ambiguity of human-machine conversation.Taking the subject-oriented human-machine dialogue system as the application carrier,the subject dictionary is imported into the natural language word segmentation module as the user-de:fined dictionary to solve the semantic ambiguity at the input side;in the dialogue management module,statistical algorithm is used to calculate the occurrence frequency of words,the words with higher frequency are used as keywords and the short text correlation is calculated with the corpus.Finally,the text with the highest correlation value is calculated and this output is fed back to the user.