| With the popularity of online shopping in life,product reviews play an increasingly important role in the user's purchase process.The discussion about reviews has always been a hot topic of research.Previous studies have analyzed the character of reviews in purchasing decisions from the perspective of the quantity,ratings,text of reviews.They also explore the credibility of reviews.Some scholars discuss reasons for consumers to adopt reviews and motives for issuing comments from the perspective of user behavior theory.However,few study explore the diffusion of reviews from a macro perspective.Reviews reflect the consumer's attention and loyalty to products.By understanding the diffusion rules of reviews,the business or online shopping platforms can better understand the trend of the user's reviews,build strategies to optimize the product's review mechanism,and contribute to consumers' understanding of the product..This article draws on the idea of innovation diffusion theory,based on the review data of 306 products on the shopping platform,and applies the symbolic regression method to provide analytical models for explaining the diffusion of product reviews.This method is an application of genetic programming algorithm based on evolutionary computation.When there are few professional experiences,it can automatically discover potential models in the data.It has been widely used in many new fields to explore intrinsic relation among variates.We first explore analytical models of online reviews diffusion.Further,Products are divided into high involvement products and low involvement products.The similarities and differences of the diffusion rules of the two types of reviews are discussed separately.The following points are finally found:1.The diffusion of reviews on the online shopping platform performs many forms.The symbolic regression method expresses these diffusion patterns by analytical models.There are polynomial models explaining the review spread with high accuracy and versatility,for example,y = a + bx + cx3-dx2.2.The classical innovation diffusion models appear in the list of reviews diffusion models which are generated by the symbolic regression method.They include the Bass model,the Fisher model and the Floyd model.Among them,the Bass model has the best coverage among the candidate models.The Fisher model is one of the three models with above-average accuracy and coverage.The classical diffusion theory of innovation can partly explain the diffusion of reviews.3.There are 4 more diffusion models for high involvement product reviews than low involvement products.The same patterns of models have stronger explanatory power and wider coverage for the diffusion of low involvement product reviews.The diffusion of high involvement product reviews have more varied forms.If one type of model is used to summarize reviews diffusion laws for high involvement products reviews,there will be more information lost than the low involvement.4 The best model for describing the diffusion of high involvement product reviews and low involvement products is the Fisher model,where high involvement product reviews peak faster.Therefore,we can find that symbolic regression method provides analytical models for describing diffusion laws of product reviews on the shopping platform and that the classical diffusion models of innovation can partly explain the reviews diffusion phenomenon.We also discuss the diffusion of product reviews with different involvement. |
PDF Full Text Request