Research On User Satisfaction Based On Online Reviews Clustering

With the rapid development of the Internet,all aspects of social life have been greatly affected.More and more people have expressed their views and opinions through the Internet,which has changed the phenomenon that original users merely considered as information receivers.Online reviews can intuitively and accurately reflect the user' s views and market conditions.Users can post the experience and feelings of the shopping process in the form of text.These reviews not only draw the attention of e-commerce platforms and merchants,but also draw potential consumers' attention.Online reviews usually include the attributes of products,merchant service content,and consumers' emotional tendencies.Enterprises can analyze the users' satisfaction of each attribute of the product through the information contained in online reviews,and formulate plans for the next development based on the industry situation.Consumers can understand the advantages and disadvantages of each attribute of the product through online reviews to make better purchase decisions and reduce risks.In the past,user satisfaction research often used questionnaires to obtain research data.There are certain deficiencies in sample size,research cost,and reliability of research results.Online reviews include not only the user's feelings about the use of the product but also the attitude after use.Users often express their satisfaction of the product through some emotional words in the review,so researching user satisfaction through online reviews not only guarantees the number and reliability of the samples,also guarantees the authenticity of the research results.Based on the review of domestic and foreign research results,this article studies user satisfaction through online reviews,researching from four aspects:feature word extraction,user satisfaction evaluation dimension construction,user satisfaction calculation,and user satisfaction analysis,which help dig up useful information in online reviews.Details as followsTo construct a user satisfaction research framework,first of all,online reviews are used as research objects to extract product feature words.These feature words can represent attribute of the product,even if a review may contain multiple features of the product.Use the word2vec tool to train the word vector model.For the trained model,use Canopy to determine the k value of the clustering number,and then use the k value as the K-means' input center number to cluster the text data,and use the clustering result as user satisfaction evaluation dimension.Then use Dependency syntax analysis to extract characteristic emotion pairs.Users usually express their satisfaction status through positive or negative emotion words matched with feature words in comments.Then calculate the weight of emotion words by extracting the dependency syntactic relationship to find that users are different dimensional satisfaction.Finally,a comparative analysis of user satisfaction in different dimensions provides suggestions for merchants to improve product quality and determine the development direction of new product.Empirical research on the determined satisfaction research framework through Jingdong Mall mobile reviews,applying the user satisfaction research method in this article to practice,crawling online reviews of users of Jingdong Mall's four mobile phones,extracting feature words from review data,determining five user satisfaction evaluation dimensions after clustering.In order to improve the reliability of the research results,this paper builds a domain sentiment dictionary on the original basic sentiment dictionary HowNet sentiment dictionary for the research field.In order to improve the accuracy of the research result,a dedicated sentiment lexicon is built for special feature words.According to the calculation results of user satisfaction,the user satisfaction of the four brands is analyzed from two perspectives.It provides advice and guidance for consumers to purchase products and businesses to develop products by using the user satisfaction quartile to analyze user satisfaction and weight from each evaluation dimension,and using Benchmarking management to compare user satisfaction in each evaluation dimension.This research theoretically enriches the research methods of user satisfaction research and combine user satisfaction research with online reviews and studies user satisfaction from feature words and compares the research results to improve the research granularity in this field.It solves the problems of coarse-grained research and the lack of scientific analysis in this field,and provides new methods and ideas for user satisfaction research.In practice,it can help consumers make better purchasing decisions based on demands,and provide a basis for businesses to optimize products and find market positioning.