Research On Text Mining Application Based On Price Classification Of Mobile Product Reviews

With the rapid development of Chinese Internet e-commerce industry,online shopping has become the main way of consumption.At the same time,it has accumulated a large number of product reviews.A large number of Chinese text comments have great commercial information value.It is of great significance in distinguishing the potential demand differences of users,improving the competitiveness of commodities,and optimizing business strategies.By learning from the theoretical experience of text mining at home and abroad,this paper attempts to establish an experimental mining method for the emotional tendencies and characteristics contained in Chinese text commodity reviews.Based on web crawler technology and natural language processing method,the research object is mobile phone product reviews in different price ranges of Jingdong Mall.Focusing on the needs of different user groups for smartphones,59,967 real Chinese text comments(unstructured data)are used to conduct a series of statistical analysis and establish mathematical models.The research object of this paper is Chinese text reviews of mobile phone products in JD.com.After preprocessing the collected review data,we perform word frequency statistics and word cloud graphs to summarize the characteristics of cognitive products.In order to analyze the sentiment tendency,randomly select comments from all the collected data to mark the sentiment tendency by manual identification,until 4000 positive and negative comments are obtained.We construct a sentiment dictionary,and use sentiment scoring to classify the sentiment of the above comment data based on the sentiment polarity of the manual annotation.Finally we calculate the classification accuracy.At the same time,we choose three text classification algorithms of support vector machine,naive Bayes and decision tree to compare the classification effect.First,we use the training set to build a classification model and apply the model to the test set.We evaluate the classifier performance through three indicators:precision,recall,and value.Finally we compare the accuracy of these classification methods.We choose the SVM classifier because it has the best classification performance.And we use it to perform sentiment analysis on all product review data in different price ranges.Then we construct LDA topic model and semantic network for feature analysis.After summarizing the real needs of users with different consumption levels for commodities,we find out the defects of the products themselves and the problems existing in the after-sale process.Based on the research conclusions of this paper,we put forward suggestions from two aspects of mobile phone production and sales and consumer purchase.First,we propose specific optimization solutions based on the characteristics of mobile phones at different price points to upgrade the user experience in an all-round way.In order to improve the quality of mobile phone sales services,we need to improve the quality of after-sales service,increase the insured price system and designate an exclusive service plan for express delivery.We need to increase the research and development of mobile phone manufacturing,solve common problems,and accelerate the speed of industrial development.Second,in order to prevent consumers from buying blindly,we need to clarify product positioning and product features for mobile phones of various price points.This will help consumers choose the most suitable product.