| With the rapid development of the information age and the improvement of people’s living standards,the demand for healthy living is getting higher,such as the concern for dental health,so the demand for electric toothbrushes is increasing.In recent years,along with the rise of the e-commerce industry,all items related to life could be purchased in the e-commerce platform.Therefore,how to find the right product for consumers in the vast amount of sales information has become an important topic.For those who buy electric toothbrushes,the product reviews could provide the best reference information.We focus on text sentiment analysis and the usefulness of reviews,in order to help consumers dig out more valuable information for purchasing electric toothbrushes.This paper crawled 10,000 electric toothbrush reviews based on Python in JD.COM.The unstructured data were processed by operations such as data cleaning,jieba word separation and text vectorization.After that,the results were compared by building sentiment dictionary and decision tree model.It was found that the results using the decision tree algorithm were better than the sentiment dictionary,and the decision tree visualization results were output to analyze consumer needs.At the same time,the LDA topic model was used to classify the topics and extract the five topics that consumers care most,which are “e?ect”,“function”,“logistics”,“appearance” and “price”.Based on these results,suggestions could be made to businesses and manufacturers to rectify the situation.Finally,four features were selected,namely “user rating”,“number of user replies to comments”,“number of days to evaluate”,“length of text”.A random forest model was built to analyze the importance of the influence of review usefulness.It was found that the feature“user rating” has little reference value for consumers,and the remain three features could be used as factors that could provide valuable information to consumers. |
PDF Full Text Request