Sentiment Analysis Based On E-commerce Product Reviews ——A Case Study Of Automatic Washing Machine In Tmall

In recent years,with the rapid development of information networks,e-commerce has also developed rapidly.The massive e-commerce product review data involves a large number of user experiences and feelings,and contains rich commercial value.This paper takes the fully automatic washing machine of Tmall Mall as an example,through data collection,data cleaning,establishment of product review index system,keyword phrase sentiment classification,and product review data analysis,to mine the value of e-commerce reviews,which is not only a potential consumer Provides reference information,as well as optimization suggestions for commodity manufacturers.First of all,this paper collects the product review data of the fully automatic washing machine industry from the Tmall mall platform through web crawlers.Due to the irregularity of the collected data,it needs to be preprocessed first.Secondly,build an e-commerce product review index system,including the design of product review indicators,the extraction of product review keywords,the extraction of product review keyword phrases,and sentiment classification.The preprocessed review data is subjected to word segmentation,and the product review index is formulated according to the word frequency statistics after word segmentation and the characteristics of the product itself;then the TF-IDF method is used to extract the index keywords from the product reviews,and then the index keywords are used to compare the index keywords.Commodity reviews are regularized and matched to obtain index keyword phrases;then some keyword phrases are manually annotated,and the keyword phrases are divided into positive and negative sentiments;finally,the labeled data set is modeled,compare the model effects of the Stacking algorithm and the LSTM algorithm,and finally choose the modified cross-entropy LSTM model to classify the sentiment of keyword phrases.Finally,through data preprocessing,index system construction,and sentiment classification of keyword phrases on e-commerce product reviews,indexed e-commerce product reviews are formed,and data analysis is performed to obtain consumers’ perceptions of various features of the products attention and satisfaction.The results show that consumers attach more importance to the appearance,volume and service experience of the product,and they are less satisfied with the appearance,volume,logistics service,installation service,price and other characteristics of the product.According to the analysis results,it provides reference opinions for potential consumers to choose products,and puts forward targeted suggestions for product manufacturers to optimize products.