A Research On The Robust Sentiment Classification Of Online Shopping Product Reviews Based On Text Mining

With the rapid development of the Internet,people's daily shopping places have gradually changed from offline physical stores to online stores.Product reviews,as the feedback of the market generated by consumers in the process of consumption,play an important role in online sales and customer shopping.In recent years,as more and more people adapt to this online shopping mode,its scale is increasing.Considering that,for online shopping users,relevant comment data may affect the final purchase decision of users,and for businesses,comment data can be used as an important basis for them to obtain business decision information and extract valuable information for consumers and businesses.Therefore,how to analyze and mine online shopping comment data,And the effective use of useful information is becoming an important research direction of sentiment analysis of commodity reviews.But it is worth noting that the data mining and analysis methods are different from the traditional methods.Online shopping user reviews are unstructured data,and the traditional data mining methods are not suitable for online shopping user reviews analysis.In view of this,the existing literature has carried out a large number of preliminary studies.However,it is worth mentioning that the existing methods are easily affected by outliers and have great limitations in practice.To solve this problem,based on the robust clustering analysis method,this paper focuses on how to apply the construction of dictionary,machine learning method and deep learning method to sentiment Tendency Classification Analysis of comment data.This paper mainly takes Huawei p40 pro and Huawei P40 mobile phones on the platform of Jingdong Mall as examples,crawls nearly 20000 comment data,and uses word cloud visualization technology and LDA topic model to conduct semantic mining analysis on the comment data,and extracts the keywords and topics of the comment data.The final analysis results show that the results of this robust method are closer to the objective reality.From the perspective of product type,Huawei p40 pro has a higher praise rate than Huawei P40,mainly reflected in the photo taking function and running speed.Huawei P40 is more popular than Huawei p40 pro because of its small size,convenient use and cost performance.From the perspective of consumer classification,members and non members focus on the same performance and function of mobile phones.According to the data collected,member consumers pay more attention to the performance,memory and system satisfaction of mobile phones,while non member consumers mainly focus on the appearance of mobile phones,the clarity of photos and the cost performance of products.From the perspective of mobile phone characteristics,the screen,cost performance,logistics,sound quality and other aspects need to be improved,especially the screen and cost performance.The appearance design,function and customer service of mobile phone have been generally recognized by users.Through further analysis of the results,this paper compares the evaluation of the basic version and the upgraded version,the needs of members and non members,analyzes whether the consumer needs are consistent with the products provided by businesses,more intuitively reflects the consumer needs and the places where the products need to be upgraded,and makes full use of the information reflected by the review data to put forward reasonable suggestions for businesses and consumers.