Analysis And Research On Network Public Opinion For The Micro-blog Hot Topic Of "Self-service Supermarket"

The rapid development of artificial intelligence(AI)makes the popularization of"self-service supermarket" to be realizable.Simple e-commerce or entity services have no longer been able to satisfy consumers' shopping experience,and the "self-service supermarket" forms a new retail shopping mode by integrating online and offline:on one hand,it saves the time for customers to queue up and pay the bills,and reduces the labor cost of the operators;On the other hand,the sellers make personalized sales plans according to the sales data,so as to provide a more intelligent service experience for consumers.In this paper,the crawler technology has been adopted to collect the textual data related with the "self-service supermarket" on the Sina micro-blog,and the development of public opinion has been supervised through the network public opinion analysis.Moreover,sentiment analysis has been applied to effectively master the sentiment orientation of subjective text,and cluster analysis is combined on this basis to help enterprises further dig the views and attitudes of netizens,thus providing decision-making support for the operation of "self-service supermarket" or even new retail mode.Generally,the sentiment analysis can be divided into two methods of sentiment lexicon based and machine learning based,which have both advantages and disadvantages:the sentiment lexicon based method is easy to operate,but it has higher requirement to the sentiment lexicon and can't identify the unregistered words;the machine learning based method solves the problem of new vocabularies,but it isolates the analysis feature term and neglects the connection between contexts.To solve above mentioned problems,new sentiment orientation classifier has been proposed in this paper on the basis of boosting idea,and research results show that:(1)On account of the boosting algorithmic thinking,weighted summation has been carried out based on the lexicon matching algorithm and machine learning algorithm,and new comprehensive classification method has been proposed,which makes the accuracy rate of evaluating indicator(0.801),recall rate(0.76)and F value(0.773)are significantly higher than the first two algorithms;(2)Through the cloud chart of words,focus of netizens on the "self-service supermarket" has been clearly understood,and the sentiment orientation has been controlled through the sentiment analysis.Finally,Text Clustering K-means algorithm has been used to further dig the public opinions and attitudes,thus better understanding the network public opinion and putting forward relevant operation strategy.(3)Generally,the netizens show a positive emotional orientation to the "self-service supermarket",but they still worry about the "quality'and "unemployment",so the enterprise and government should correctly guide the consumers' public opinion.On one hand,they shall make pertinence analysis of the focus of "self-service supermarket" in the Micro-blog content operation,do a good job on public opinion guidance;On the other hand,they shall show the "self-service supermarket" to public in a more lively and quick way through multi channel cooperation,such as WeChat,Micro-blog and live broadcast.(4)With "self-service supermarket" as an example,it can be promoted in the new retail industry.By studying the full channels of new retail business and characteristics of intelligent operation,early warning mechanism of network public opinion could be used to correctly guide and supervise the public opinion,and ensure its healthy development.