Research On The Improvement Of Domestic TV Brands

The development of the Internet provides users with a window for information interaction and emotional expression,at the same time online comments posted by users represent their actual feelings.On the one hand,these online reviews provide guidance for other users to purchase,on the other hand,they also allow enterprises to clarify their own advantages and disadvantages,which is convenient for product iterative upgrades.However,with the development of e-commerce platforms,the amount of online review data has exploded.Faced with mixed review data,it is difficult for merchants and users to make accurate judgments.Therefore,how to make better use of the massive amount of review data to guide companies to improve their products and enhance brand competitiveness is a worthy study.This thesis selects the online review data of JD.com,the largest domestic electrical appliance retail e-commerce platform,as an example.The main research work is as follows:First,according to the sales situation of the color TV industry in the first half of 2021,this thesis selects LG Electronics,which ranks in the first echelon of foreign brands,and TCL,Xiaomi and Huawei,which are in the first and second echelons of domestic brands,respectively.Use web crawler to obtain 13028 pieces of data,and preprocess them,including removing stop words,word segmentation,deduplication,etc.,and then use word frequency statistical analysis,word cloud graph and network relationship collinear graph to understand the user’s demand points.The results show that,the user’s focus is mainly on the appearance of the product,price,audio and video effects,service and so on.Second,sentiment analysis research.This thesis compares traditional machine learning algorithms and deep learning algorithms,and finds that the BiLSTM algorithm is better for sentiment analysis,and its performance is 6%higher than that of traditional machine learning methods.Finally,the sentiment analysis of all the review data shows that the foreign brand LG’s praise rate is over 90%,while domestic brands’ praise rate is 50%-60%,indicating that domestic brands need to further improve their products to enhance the user experience.Third,the usefulness analysis of the review text.This thesis analyzes the length of the review text,the number of replies,the product relevance,the difference between the consumer’s purchase time and the comment time,emotional polarity,and consumer membership information.Taking them all into consideration,then construct four types of evaluation indicators.At the same time,compare the classification results of Random Forest,SVM and XGBoost algorithms,and the study finds that XGBoost is more effective as a prediction model.Fourth,establish the LDA theme model,and obtain the advantages and disadvantages of each product through mining these online reviews.Finally,according to the previous analysis results,put forward some improvement suggestions for domestic color TV brands,aiming to further enhance the influence of domestic brands.