Research On Electronic Word-of-Mouth Information Diffusion Model In Online Social Networks

Social media has become an important channel for rapid dissemination of information.As an important reference for customers' shopping decision,eWOM is crucial for business performance and brand image.The diffusion of positive eWOM can actively promote product sales while the spread of negative eWOM may bring great challenge to the brand reputation.In the context of social media,the complexity of social network structure and the diversity of influencing factors of the diffusion bring new opportunities and challenges to the study of eWOM diffusion.Therefore,this paper conducts analysis on the eWOM diffusion model from the perspectives of customer,network structure and eWOM sentiment.The main research objectives are as follows:1)In view of the phenomenon of consumer interest shift in online word-of-mouth communication of short life cycle products,based on epidemic model,an eWOM diffusion model considering individual interest shift and social trust is constructed.In the model,an uninterest state is added to stress consumer interest shift behavior,and community structure is applied to describe the social relationship and the trust level between consumers.The simulation and numerical analysis indicate that the interest shift behavior can directly influence the outbreak size of the eWOM information dissemination and therefore influence the diffusion speed.Further,when the modularity of the network is higher or the community clustering coefficient is lower,individual interest shift behavior will have a heavier effect on the spread scope.In addition,the trust level between customers has a significant effect on the eWOM diffusion speed and scope.Moreover,compared with the improvement of the trust between customers inside the same community,the improvement of trust between consumers in different communities can promote eWOM diffusion at an earlier stage,and the impact is more significant.2)In response to the time-varying characteristics of consumer social networks in the spread of long-life cycle products online word-of-mouth,a dynamic network structure description method based on individual mobility is proposed.On this basis,an eWOM diffusion model under dynamic network structure is constructed.By solving the basic reproduction number of the model,it is found that the flow of consumers between different communities reduces the eWOM diffusion threshold and is conducive to the beginning of communication.Simulation and numerical analysis show that the individual flow rate and community attractiveness significantly affect the eWOM diffusion process,especially in the early stage of the diffusion.Consumers' aggregation effects caused by differences in mobility and community attractiveness can significantly increase the local spread of dominant communities and thus enhance the overall spread rate.In addition,in networks with strong community structure,the promotion effect of individual mobility is more significant.However,it is also found that social mobility does not always effectively accelerate the spread of eWOM.When all communities have the same attractiveness,raising or lowering the flow rate will no longer promote or suppress the eWOM diffusion.3)In view of the phenomenon of eWOM sentiment propagation,on the basis of the eWOM diffusion model,consumers are characterized by the state of diffusion and eWOM sentiment value,and an eWOM sentiment diffusion model considering the business intervention is constructed,which incorporates the spread of sentiment into the process of eWOM diffusion.The simulation indicates that the positive and negative ratio of the initial eWOM and the sentiment intensity can significantly affect the process of eWOM sentiment diffusion.Sentiment with high intensity is more infectious.In addition,the consumer's ability to perceive the overall sentiment information can also effectively affect the eWOM diffusion process,A higher average perception level leads to a less impact of extreme comments.Furthermore,it is found that the initial stage of eWOM diffusion is the golden time for enterprises to take actions.At this stage,same investment can bring significant suppress effect on the diffusion of negative eWOM,and the sooner the risk is perceived,the stronger the effect is.4)Sina microblog and Samsung note8 mobile phone are taken as examples to analyze the diffusion process of eWOM information and eWOM sentiment.Through the actual data analysis,it is found that the diffusion process on microblog has obvious " outbreak-asymptotic effect ",that is,there is a rapid outbreak at the initial stage,followed by the slow continuous diffusion process.The eWOM diffusion model based on individual interest shift proposed in this paper can effectively describe this process.Furthermore,a sentiment classification method based on topics sentiment similarity is proposed,and the positive and negative polarity of online reviews is automatically identified.The eWOM sentiment diffusion process of Samsung note8 mobile phone is analyzed and the validity of the eWOM sentiment diffusion model is verified.This thesis is closed with a summary followed by a list of topics for future research.