Research On User Satisfaction Of Small New Energy Vehicles Based On Sentiment Analysi

At present,more and more users refer to online reviews to make purchasing decisions,and the importance of online reviews is becoming increasingly prominent.Online review data is characterized by a large number,clear views,short pages,and casual expression.Manually browsing through massive online reviews is not only time-consuming,laborious,but also inefficient.In contrast,emotion recognition obtains information quickly and efficiently,and can mine out user emotional characteristics,so as to grasp user satisfaction and provide a basis for relevant corporate decisions.How to more effectively mine valuable information from user reviews has become an important research topic.Based on deep learning technology,this thesis constructs a sentiment analysis model for online reviews of small new energy vehicles,explores the influencing factors of user satisfaction of small new energy vehicles,and proposes a way to obtain user demand based on Internet review data.Firstly,the Python crawler tool is used to grab the word-of-mouth review data of the i Card Auto website,and the final effective comment sample is obtained through data cleaning,text word segmentation and stop word removal,which is used as the data source for sentiment analysis.Then,the new energy vehicle product attribute thesaurus and emotion dictionary are constructed,and the attention mechanism is introduced by combining the CNN model,BiLSTM model and BiLSTM-CNN model,and a small NEV online comment sentiment classification model based on BiLSTM-CNN-Attention is formed.The advantages of BiLSTMCNN-Attention model in the online review sentiment classification of small new energy vehicles are verified by experiments.Finally,with the help of XGBoost model,the importance of high-frequency attributes affecting user satisfaction of small new energy vehicles is calculated,and the user’s satisfaction with each high-frequency attribute is measured from the importance of high-frequency attributes and its positive emotional intensity,and the influence of small new energy vehicle attributes on user satisfaction is explored,so as to clarify the improvement direction of automotive products.It is found that the introduction of attention mechanism in the BiLSTM-CNN model can reduce information loss and information redundancy in the process of feature vector extraction,and further improve the accuracy of model sentiment classification.Based on the analysis of user satisfaction influencing factors of XGBoost,it is found that among the attributes of small new energy vehicles,the five attributes with the highest user satisfaction index are power,appearance,fuel consumption,cost performance and interior,and the five attributes with low customer satisfaction are space,displacement,technology,endurance and safety.Therefore,enterprises need to improve the space,displacement,technology,endurance,safety,etc.of small new energy vehicles to meet user needs and improve user satisfaction.