Research On The Disambiguation Of Natural Language Dialogue Based On Statistics And Rules And Its Application In Product Design

With the development of information technology,especially the breakthrough of Artifical Intelligence(AI),it has injected vitality into the new development of the machinery industry.The natural language understanding(NLU)is applied to the acquisition of mechanical product design requirments,demand analysis,mechanism design,assembly implementation,and mechanical design.In various aspects such as software assistance,physical labor of practioners is further reduced and efficiency of artificial design is improved.Due to the existence of the actual environment,when using natural language communication,omitting part of environmental information in speech does not produce ambiguity;the same word with different meanings in different actual environments can reduce its ambiguity according to the actual environment;in the actual environment When object properties are brought into understanding,they complement and improve the understanding of information based on environmental relevance.The Actual Environment-Based Dialogue ellipsis ambiguity processing model proposed in this paper is to bring actual environmental information,actual environmental knowledge information,dialogue,and context understanding into natural language dialogue and understanding process,reduce environmental information omitted,dialogue information is blurred,context refers to ambiguity caused by unknown understanding.Firstly,according to the mapping relationship between the natural language word segmentation result and the Actual Environment Dictionary Database to calculate the language information and the Actual Environment,the Actual Environment involved in the language is determined.The model is used to sort the Actual Environment,and then the Actual Environment Fact Library is used to supplement the perfect environment.Information,and expand related information based on the Actual Environment Knowledge Base,use contextual information to supplement fuzzy information;Secondly,based on the Actual Environment Dictionary,the secondary word segmentation of the supplemented language information is used to improve the fit between the segmentation result and the Actual Environment,and reduce the Segmentation ambiguity;Again using natural language semantic analysis,parsing the semantic collocation relation,giving understanding results,presenting the executable results to the user through the output;Finally determining the questioning information to obtain accurate information of the fuzzy information or problem.This design,let the NLU is not just limited to the understanding of the language,let the natural language information with the Actual Environment resource scheduling,method calls,such as interaction,improve the satisfaction degree of the users of the Products,give the related technical staff to help.We apply this model to the mechanical product design process of Pro/E software.We mainly implement assembly design and adjustment of parts,call methods in Pro/E software environment,and help users complete benchmark selection,adjustment of constraints,etc.,to achieve natural language the conversion to the Actual Environment was given to the technical personnel in the relevant fields to improve the user's satisfaction with the product design.