Research On Sentiment Analysis Based On E-commerce Product Reviews

Due to the fierce competition of e-commerce platforms,it is very necessary to know more about the voices of consumers.E-commerce platform users will post comments after purchasing products,and these comments reflect users' experience and real feelings.Sentiment analysis based on users' textual comment data can provide more detailed reference for product to improve itself,and can also help enterprize optimize marketing strategies.By providing suggestions to e-commerce platforms,it is possible to improve the user's loyalty to the product and the user's activity within the platform.First of all,this paper preprocesses the data of Huawei P50 pro mobile phone users' comments on the Jingdong platform,and establishes a three-way decision combination classification model based on support vector machine(SVM),k-nearest neighbor(k NN),and naive Bayes(NB)algorithm.Compared with other singleclassifier models,the model has a significant advantage in classification performance.Secondly,according to the crawled users' reviews of five different mobile phone brands,we conduct sentiment analysis and topic extraction research.According to the established three decision-making combination model,the specific number and proportion of positive and negative comments are obtained,and the LDA topic model is established according to the positive and negative comments of users after classification.Through the extraction results,it is more intuitive to understand the positive and negative experiences of users on different mobile phones.Finally,based on Huawei P50 pro mobile phone user reviews,we combine with 1V1/1Vn emotionfeature rule base,dependency syntax analysis and other methods,we extract featuresentiment word pairs from explicit feature emotional sentences,implicit feature sentences and implicit emotional sentences,which help us quantify users' emotional attitudes towards product features,and draw the corresponding conclusions.