User Demand Mining Analysis Of X Brand Mobile Phone Based On Online Reviews

At present,user demand is the basis of product improvement and innovation.However,the traditional method of obtaining user demand can not meet the requirements of enterprises.Online reviews include the user's evaluation of each attribute of the product,which can help enterprises understand the user's satisfaction with the product,so as to provide reference information for product improvement.In recent years,online reviews have become an important source of information for enterprise' innovation and improvement due to the many advantages of obtaining user demand from online reviews.Therefore,research on how to analyze user needs based on online reviews is of great significance for product improvement.In terms of user demand analysis based on online reviews,current research only analyzes from a single static or dynamic perspective,ignoring the comprehensive effect of the two perspectives on product design improvement.The product problems found by static analysis and dynamic analysis are different,so the improvement strategies proposed are different.This paper analyzes user demand from two perspectives: static and dynamic analysis,aiming to help companies improve the quality and efficiency of product improvement.The main research contents of this paper include the following aspects:First,the construction of feature lexicon in the field of mobile phone.In this paper,we built an initial lexicon based on the product parameters of mobile phone,and used word2 vec to expand the lexicon.The final feature lexicon contains feature-attribute double-layer structure.Second,the construction of online review usefulness classification model.On the basis of literature review,this paper defined the usefulness of online reviews and selected impact indicators accordingly.Then this paper compared the classification effect of BP neural network,SVM,random forest and decision tree,and selected the better classification algorithm for online review usefulness screen.Third,user demand analysis.Firstly,dependency syntax analysis was used to extract four tuples of <attribute,negative word,degree word,emotional word>.Secondly,based on the sentiment lexicon,the four tuples were calculated to get the attribute sentiment value.Thirdly,the static user demand of the product was analyzed.Finally,we analyzed the changes of user demand in the two periods and conduct dynamic analysis.And we integrated the two perspectives for multiangle analysis,classified user demand,and proposed product improvement strategies.The results show that: First,the accuracy of BP neural network is 87.5%,which is better than the other three algorithms.Second,four types of user demand can be obtained based on the user demand mining model based on online reviews.Each type of demands reflects different problems.Enterprises can make targeted improvement strategies.Third,Through the comparison of singleangle analysis and multi-angle analysis,it can be found that multi-angle analysis is more conducive for enterprises to find out product problems and whether the product improvement direction meet the development direction of user demand,and promote to improve the quality and efficiency of product improvement.