The Analysis And Countermeasure Research Of Negative Brand Affect Based On Weibo Comments

Since twenty-first Century,with the rapid development of economy and science and technology,new products and brands emerge one after another.In the fierce market competition environment,the competitions of products gradually turn into the competitions of brands,which make the brands become one of the most popular keywords.The unique charm of the brand is reflected in its ability that not only meets the needs,but also brings emotional experience.The effective establishment of brand affect plays a key role in avoiding the homogenization of products,forming the brand value,building the core value of brand and enhancing brand equity,is also very important for whether the enterprise can stand out in the fierce market competition.At present,the research of the traditional brand affect is based on the questionnaires,interviews and other subjective forms,which have less convincing data,lead to enterprises cannot get really valuable information,and this is one of the key issues that have been restricting the accuracy of brand affect research for a long time.In recent years,with the development of the Internet technology,Sina Weibo,as a representative of the online social media,has become the main carrier of people's interaction and communication.Therefore,the brand official Weibo has also become an important battlefield of brand operations,which contains a large number of comments on the brand affects' tendencies,and reflects people's judgments on the quality of products and services.Based on the Weibo to acquire massive,real-time and efficient user comments,synthesizing theoretical results of mature brand effect research,let the brand and Weibo to achieve the deep combination of business innovation and information technology,and which will greatly bring great impetus of shaping brand affect and promoting business value.In addition,relative to the positive brand effect in the comments,the negative brand affect is easier to spread and transmit,therefore,enterprises need to focus on its management and control;at the same time,people's reactions are more intense and honest,so its content is more accurate and reliable.It can be seen that the research on negative brand effect has stronger practical significance.In this paper,based on the study of the traditional brand affect,taking Huawei brand as an example,based on Weibo comments,put forward the method of data mining to find theaffective tendencies of brand,and to identify positive and negative comments on the brand affects.Find and analyze problems on the negative brand affect in the negative comment text,and put forward the corresponding countermeasures and suggestions,so as to control and manage the negative brand affect,adjust the company's brand strategy timely and reasonably.The main contents of this paper include:First,based on Weibo comments,establish a sentiment classification model of brand affect.First of all,take the use of the web crawler technology to get Weibo comments,and select 7000 user randomly,to carry on the statistical analysis about comment time,individual basic information;secondly,complete the data pre-processing task.Carry out data pre-processing with the initial comments,such as removing weight,removing noise and segmenting words.At the same time,based on deep learning method,on the basis of a large set of comment data sets(about 230 thousand comments),train the text feature expression model--word2 vec model.Finally,select 15000 comments randomly,use manual emotion polarity annotation,classify the comments.And separate 15000 comments into the training data sets and test data sets,respectively.Use three machine learning algorithms to train the sentiment classification of brand affect,which are Deep Belief Network,Support Vector Machine and Random Forest,respectively.Use the ten-fold cross-validation method to examine the accuracy of the classifier results,through the comparison of the accuracy of three different classifier models,select the Random Forest model of the 500 dimension as the optimal dimension and optimal algorithm sentiment classification model.Second,based on the established classification model,this paper makes an empirical study on the negative brand affect of Huawei's official Weibo comments.After analyzing the result of the classifier on the brand affect,get the negative comments of brand affect.From four aspects,which are the product type,product attribute,service and marketing and consumer related emotion and action expression,do word frequency statistics,analyze the monthly evolution trend of negative ratio to text and high frequency words,from the product,service,brand recognition these three aspects to find and analyze the problems.Third,aiming at the problems of negative brand affect,to establish a deep emotional relationship between the brand and consumers,from the aspects of product,service,brand recognition to provide measures and suggestions.Thus,by doing so,provide a new way and auseful channel for brand affect research,and is also conductive to implement a better strategic adjustment,establish a good brand image,enhance the core competitiveness of the brand.