Research On Semi-structured Text Push Technology And Application

The complex product has the characteristics of complicated structure,knowledge-intensive and long cycle of development.Reusing existing design knowledge will make contribution to shorten the development cycle and improve design quality.Currently,case-based and rule-based knowledge reusing technology are much more mature,while for the lacking of structure,the research on the reusing technology of patents and other semi-structured text in conceptual design process of construction machinery product still have shortcomings,with the accumulation of design knowledge,its use efficiency will decline sharply.Therefore,in order to improve the reuse of semi-structured text and realize its initiative and timely push in the conceptual design process of construction machinery product,the push technology of semi-structured text and its application system were studied.Firstly,the knowledge type and requirement of the conceptual design process of construction machinery product was analyzed.The classification and modeling method of product conceptual design knowledge was proposed.And then,the static and dynamic knowledge model of product were established.After that,based on the dynamic knowledge,all kinds of knowledge were organized.Then,according to the problem of low efficiency in rough classification of semistructured text,in order to realize its fine-grained classification,a semi-structured text classification and mining method was firstly proposed,the basic process of it was given,and a new multi-label K-nearest neighbor classification method was proposed,which indirect considered the relevance between different labels,the relevant parameters affecting the classification performance were estimated to narrow the search range and improve the accuracy of its retrieval.After that,in order to solve the problem of low retrieval efficiency of semi-structured text in the current retrieval mode,and the problem of repeated retrieval due to the lack of summary of decision information,the knowledge push technology was studied and the knowledge push method which was integrated user similarity and project similarity was presented.By using this method,after the processing of the fine-grained classification of semi-structured text,the appropriate suggestion list of knowledge will be generated,which may help to realize the initiative push of semi-structured text and improve the efficiency of filtering and screening the usefulness of design knowledge.Finally,based on the actual system requirement of Weichai Power Co.,Ltd.,the web-based green design support system of engineering machinery powertrain was developed from the enterprise knowledge demand.The architecture and operation mechanism of the system were discussed in detail,and the real product design task was utilized to introduce the functions.Based on this,the correlation algorithms and the design support system were verified.