Real User Experience Recognition Based On Automobile UGC

The Internet age has made it infiltrate into various fields,and more and more excellent vertical websites have also emerged in the automotive field.Correspondingly,there are also a large number of user-generated content,namely UGC(User Generated Content).However,the UGC of these websites is not all true.It causes a certain amount of distress to a large number of users and those who need this data mining information.At the same time,as UGC becomes more and more an important channel for people to obtain information sources,if user experience and needs can be unearthed from it,it will provide certain reference opinions for the improvement of automotive products.This paper analyzes and excavates the car experience of real users based on the car UGC.The main research content includes the following aspects:(1)A method to identify abnormal users based on the content of the comment text is proposed.According to the naive Bayes algorithm,the support vector machine algorithm and the long and short-term memory neural network,three models for identifying abnormal users were constructed,and verified based on actual data,and the model with higher recognition accuracy was selected,and the model was used to identify and classify the text data needed in this article.(2)A method for identifying entity words and opinion words in review texts for the automotive field is proposed.According to the existing language features in the review text,several feature templates are designed,and the conditional random field is used to experiment to select the best ones.Using the theory of deep learning,a model for identifying entity words and opinion words is constructed,and the recognition effect of the model is verified by experiments.(3)A set of methods to analyze user experience based on automobile UGC is proposed,real user needs are explored,and the experience of positive abnormal users is compared and analyzed.Through the recognition of the entity words and opinion words in the user review text,the user’s opinions on automobile products are mined,and the sentiment value of each user review is scored to perform sentiment analysis on it,and the major is calculated according to the sentiment score of the review.The average sentiment value of the attribute feature.Combining the relevant theories of the quadrant graph model and improving it according to the characteristics of the data in this article,the user’s satisfaction and attention are calculated.Created the "satisfaction-attention" four-part graph model to realize the mining of real user needs,and used the methods of abnormal user identification and user experience analysis to compare and analyze the experience of positive abnormal users and explore the differences.