Research On Online Reviews Perceived Usefulness Evaluation And Application Based On Text Mining

In recent years,with the rapid development of Web 2.0 technology,online shopping as an important form of e-commerce has been widely accepted and recognized by people,has become an important way of daily consumption.The obvious difference between online shopping and traditional entity consumption is that the purchased goods can only be displayed to consumers through the way that the network can present,such as text,image,sound and so on,while the intuitive perception and trial experience of the goods are greatly limited,which makes the online word-of-mouth spread on the Internet platform highlight its tremendous influence.As a form of online word-of-mouth extension,consumers' online reviews of selected products have been developed rapidly.Online reviews are profoundly changing consumer behavior patterns,corporate profitability and marketing strategies.According to the survey data,more than half of online consumers rely on online reviews for online shopping,and 98% of consumers browse online reviews before ordering.Product suppliers or retailers can not only attract more consumers to visit their websites and help them find suitable products,but also rely on data mining and other technical means to understand consumers' shopping feelings,experiences and habits,and ultimately achieve the goal of product promotion and value growth.Online reviews have become an indispensable new element in the marketing strategy of enterprises.However,the increasing number of online reviews also brings challenges to online shopping.The generation of a large number of online reviews on e-commerce platforms every day results in information overload.It is difficult to identify and select reliable online reviews data to mine and analyze the perceived usefulness of consumers' online reviews.To solve this problem,it is of great practical significance to study the perceived usefulness evaluation of consumers' online reviews and the prediction of sales volume of online reviews based on perceived usefulness evaluation,whether in the construction of merchant production or sales enterprises or website commodity evaluation system.Based on Consumer Behavior Theory(TRA,TPB,TAM,TAM2),Zeithaml's Customer Perceived Value Theory,Information Dissemination Theory(Hofland Communication Persuasion Theory,ELM Theory),this paper studies the process andmethod of online review text mining analysis,studies the evaluation method of consumer online review perceived usefulness,and studies prediction of sales volume based on online reviews perceived usefulness evaluation are to help e-commerce enterprises(product manufacturers and operators as well as website platforms)truly understand and grasp the dynamic needs of customers and their satisfaction with product quality and service,provide reliable decision-making basis and increase the stickiness of online shopping.The main work of this paper is embodied in the following four aspects:(1)To explore the principles and methods of text mining based on semantic dictionary,and analyze its effectiveness.First,the design idea and process based on semantic dictionary are proposed.Secondly,this paper discusses the principle and method of extracting pairs of < feature words,opinion words> in online reviews.Then it discusses the principle and method of measuring emotional polarity and calculating emotional intensity of online comment based on < feature words,opinion words > pairs.It mainly analyzes the derivation process and implementation details of extracting feature words and opinion words,and expounds the online comment situation.The realization steps of polarity measurement and intensity calculation.Finally,through the analysis of the application results,it is found that the evaluation method based on semantic dictionary is superior to other machine learning methods in the accuracy of feature word classification.(2)Construction of online perceived usefulness evaluation model.Aiming at the important influencing factors of perceived usefulness of online reviews,this paper discusses the perceived usefulness of online reviews from the three dimensions of perceived usefulness of external features,perceived usefulness of internal features,credibility of reviewers.Firstly,the theoretical model of online review perceived usefulness evaluation is constructed and the evaluation index is analyzed.Based on this analysis,the evaluation index system of online review perceived usefulness is constructed.Then,the original data of Amazon's online reviews in China is captured,and the sample data of online Reviews' perceived usefulness evaluation index is obtained by text mining,cleaning and pretreatment.The validation analysis based on SEM model was carried out from three dimensions: external feature perception usefulness,internal feature perception usefulness,reviewer credibility.The significance of the impact of onlinereview perception usefulness evaluation index on the model was tested,and the rationality and robustness of SEM model were also tested(3)Application research of online reviews perception usefulness evaluation model.Aiming at the acquisition and processing of online review perceived usefulness evaluation index data,the significance test of evaluation model and the evaluation method of online review perceived usefulness,the process of online review perceived usefulness evaluation is designed.The grey comprehensive evaluation method is used to randomly select the sample records of online reviews and evaluate their perceived usefulness.(4)Online sales volume prediction based on perceived usefulness of online reviews evaluation.Firstly,the impact of perceived usefulness of online reviews on product sales is analyzed,and a product sales forecasting model based on online reviews is constructed.Then,a theoretical analysis is made for the selection of the measurement index of the prediction model.Then,on the significance test method of prediction model,a SVM regression model based on support vector machine is constructed to solve the problem that the multivariate linear regression model can not predict the large sample nonlinear data accurately.Finally,through the crawling software to collect a variety of well-known brands of mobile phone related raw data,data cleaning,pre-processing and online review text mining,statistical collation of applied forecasting sample data,classical multiple linear regression and support vector machine regression application analysis.Based on the analysis results,the prediction performances of the two models are compared,and relevant conclusions are drawn.This paper examines the applicability of consumer online review perceived usefulness evaluation model and online review product sales forecasting model based on perceived usefulness evaluation in the field of network consumption through application analysis.It provides a new perspective for online review related research in this field,and also provides a new perspective for e-commerce websites and their online review system.It provides constructive practical reference for improvement and optimization.