Text Mining For Online Purchase Of Comment Data

With the development of technology and the popularity of mobile phones,online shop-ping has penetrated into all aspects of people's lives,and the number of people that have done online shopping is increasing.There will inevitably be a large number of commodity comment data onto this process.It's very difficult to find effective information from a large amount of commentary data by manual reading.Therefore,the paper solves this problem with text mining technology.First of all,in order to obtain effective information on the comment data by using text mining technology,the paper uses Python to write a crawler program,and obtains comment data onto a refrigerator from Jingdong Mall,Tmall Mall and Suning respectively.In order to better to perform emotional classification and topic analysis on the review data,the pa-per preprocesses the obtained comment data and obtains standardized comment data.The normalized comment text is segmented and visually displayed in the form of word cloud pictures.Secondly,after the standardized text is obtained,the emotional classification method based on emotional dictionary and the improved machine learning method is used to clas-sify the emotional data,and then the accuracy index is used to compare the effect of the two emotional classification methods.By comparison,it's found that the improved machine learning method has higher emotional classification accuracy and better effect.Therefore,the emotion classification results of the improved machine learning method are selected to construct semantic networks of positive and negative comments on three electronic com-merce platforms respectively,to reproduce the relationship between comments and word segmentation,and to visually display the opinions in user comments.Finally,the topic analysis technology is used to model and analyze the positive and negative reviews of the three electronic commerce platforms respectively.The potential topics are found and summarized from the comment data,and the commodity characteristics and e-commerce platform characteristics contained in the comment sentences are mined by combining word frequency and semantic network.On the one hand,it can assist users to purchase;on the other hand,it can help commodity manufacturers and business platforms to understand users'demands and advantages and disadvantages of commodities,so as to make targeted improvements.