Study Of The Case-matching Model Supporting The Cars' Maintenance With Question-answering Method

The difficulty in obtaining the case data of vehicle maintenance,the low efficiency and the high demand of fault diagnosis are the main problems faced by service consultants and car owners in the field of vehicle maintenance.The service synergy system built by “the ASP/SaaS-based manufacturing industry value chain collaborative platform” and its maintenance data archives can support the initial judgment on vehicle maintenance for service consultant,but the platform-based diagnosis and the consulting of maintenance project also exist some questions,such as: 1)without Targeted and convenient service;2)service information is not open and service quality is easily influenced by human factors during the maintenance process;3)service couldn't understand the semantic information of the maintenance query;4)there are redundant information within too many provided query-results.There is an urgent need to establish neutral,unobstructed and transparent information channel based on third-party,in order to assist vehicle owners and serviceconsultants to complete the initial judgment of vehicle breakdown,acquisition of the maintenance knowledge,and understanding maintenance technology independently and independently.Thus,this paper's research based on thousands of car companies' maintenance date accumulated by the "ASP/SaaS-based manufacturing industry value chain collaboration platform" which is belonging to the national key research and development program "Distributed Resource Giant System and Resource Synergy Theory"(Projection No: 2017YFB1400301)in the past 10 years,and the big data of the car' maintenance cases are the base,this paper proposes a third-party question-answering system solution supporting the match of similar maintenance cases,then,the key requirements of the automatic question-answering system are analyzed.And at last completing the functional design and overall of the automatic question answering system.The paper focuses on the three core key points to support system implementation: the information analysis of the text content,semantic understanding,and the extraction for the program matching result.Firstly,accomplishing the NLP-based extraction of keywords and phrase expansion according to the data characteristics of car's maintenance cases,which aims at completing the initial analysis of user's query intent;Secondly,In order to support semantic understanding and achieve similar cases,proposing the improved algorithm based on the fusion of words and definition of terms' on the base of words' vectorization by the word2 vec model and the design of answer-sorting method by BM25 algorithm.Thirdly,a case extraction method based on continuous language model is proposed,which eliminates the information redundancy of question answers.At last,the effectiveness of the proposed method is verified by experiments.