Research On Data Mining Of Online Reviews Of E-commerce Products Based On Sentiment Analysis Technology

The rapid development of e-commerce has changed people's shopping patterns.With the rnse of online shopping,many e-commerce platforms have emerged in China.In that case,e-commerce platforms and merchants have gained a lot of development opportunities,but as the market competition continues to intensify and technology continues to develop,e-commerce platforms and businesses have also encountered many challenges.For the e-commerce platform,their goal is:a large number of merchants settle in and a large number of user registration;For merchants,their goal is to expect more and more sales of goods.In order to achieve these goals,in addition to ensuring reliable product quality and reasonable price,it is also necessary to pay attention to the consumer's shopping experience and listen to their voices.Most e-commerce platforms now have corresponding consumer online commenting systems.Successful users can post their opinions and feelings within a certain period of time.These contents are increasingly valued by merchants and platforms.This provides important intelligence data for their next development and improvement.In this paper,we researched the relevant comments and found that some comments expressed consumers' satisfaction with this shopping;Some of the comments reflect consumer criticism and are not satisfied with the purchase;There are also some online comments that we can't judge whether consumers are satisfied with the purchase.It is a neutral mood.Therefore,we conduct research and analysis on the review text data through sentiment analysis technology,and mine valuable information to provide intelligence data for merchants and platforms.This article selects two infant milk powders on the two platforms of Tmall International and JD Global to conduct online review data crawling.Then preprocess the raw data:Text deduplication and mechanical compression,Chinese word segmentation,part-of-speech tagging,removal of stop-words and English characters,etc.Then construct the relevant sentiment dictionary and make the emotional judgment based on the relevant emotional dictionary based on the pre-processed review data.Divide the comment text data into three parts:positive,negative and neutral,and construct a network semantic graph to realize data visualization.Finally,the logistics service elements in the comment text are extracted,and the relationship between consumers' willingness to purchase and the factors are explored through factor analysis and questionnaire survey,and the relevant regression models are constructed.