Research On User Experience Evaluation Of Automobile Products Based On Large-scale Text Mining

The popularity of automotive products and the rapid development of modern information and communication technologies have spawned a large number of forums,vertical websites and communities for the automotive sector.Vertical websites of the automobile field contains the word-of-mouth and comment contents published by users in the process of vehicle browsing,selecting,buying and using,which reflects the user’s subjective and comprehensive evaluation of the product based on their real experiences.These evaluation contents not only directly reflect the user’s demand for the purchased products,but also indirectly reflect the defects or pain points of the products.Therefore,the analysis of user evaluation information in the automotive field can not only quickly locate the satisfaction and complaints of the product,but also help the company to improve the design,and also help the consumer to make reasonable consumption decisions,which has important practical significance.This study adopts the text mining method based on machine learning,and takes the user evaluation and word-of-mouth content of mainstream car vertical website as the analysis object,and systematically studies the knowledge base construction,product defect automatic recognition and user subjective experience evaluation of automobile products.The main work of this research mainly includes the following aspects:(1)A construction method of knowledge base for the automotive industry is proposed.Firstly,based on the product vocabulary of the automotive field,combined with the new words homophonic words,and typos in the users’ presentation in the automotive field,the key dictionary in the automotive field is built,which is more suitable for the text content and expression characteristics of the automotive field;secondly,based on core concepts of the automotive industry and the relationship between concepts,for the category,ontology structure,performance indicators and usage scenarios of automotive products,an automotive product ontology knowledge base is built in terms of text mining,which is comprehensive and systematic and is conducive to analyze and extract Internet text.(2)A fine-grained method for subjective experience evaluation of users is proposed.Based on the knowledge base of the automobile field,the comment text is divided into product features,and the Word2 vec tool is used to obtain the sentence vector of the comment text.Based on the machine learning model,the sentiment analysis result of the text is obtained.The test results show that the method has a higher precision,recall and F1 values.(3)A user experience evaluation model for multi-source data and text mining is proposed.Using the established knowledge base of automotive products and user subjective evaluation methods,combined with the characteristics of Internet commentary,a model to measure the degree of user experience was constructed.Mainstream compact SUV models were tested by user subjective experience in the case analysis,and the trend of subjective evaluation of users in competing models was studied,and the experimental results were explained.