| The automobile industry is a strategic industry of the national economy which plays an irreplaceable role in accelerating the process of industrialization,promoting the innovative development of manufacturing,increasing employment,and promoting consumption upgrades.In recent years,in the context of the slowdown in the growth of the auto market,declining demand,and intensified competition,the problem of overcapacity among automobile companies has gradually emerged.In addition,with the ever-increasing demand of people,consumers have put forward higher and higher requirements for automotive products.In order to gain a dominant position in the fierce competition,automobile companies should accurately predict the future demand in the market to rationally plan production capacity and formulate production plans.They should fully understand consumer requirement,clarify product design priorities and discover product defects,and then improve products Market competitiveness.Online reviews have become an important source of information for consumers to make buying decisions and companies to mine customer requirements.Based on this,this article conducts the following three aspects of research: firstly,feature-sentiment word pairs are joint extracted from reviews.After that,a domain sentiment lexicon is constructed by modified method and applied in online reviews’ multi-aspect sentiment analysis.Secondly,a novel model is proposed that takes online reviews and national macroeconomic indicators into consideration to forecast sales.In this method,the prospect theory and the attribution theory are introduced to quantified the influence of reviews sentiment on sales.Finally,an innovative KANO conversion method is developed to classify customer requirements.Here,the percentage of users’ satisfaction and dissatisfaction with product features are calculated as the basis for customer requirements classification.The research result shows that: the accuracy of multi-aspect sentiment analysis can be effectively improved by a domain sentiment dictionary that constructed through feature-sentiment word pairs joint extraction and modified method;compared with only using historical sales,adding the index of perceive emotion of reviews and macroeconomic indicators to model can effectively improve the accuracy of forecast;by extracting the customers’ sentiment orientation of product features in reviews and constructing KANO model,it can effectively mine customer requirement and guide product design. |
PDF Full Text Request