Research On Information Diffusion Mode Of Consumers' Self-organization And The Process Control Method

Information diffusion on social network platforms now has become the main way in human's daily communication,whose considerable values have received much concern in both academic and industrial circles.However,the previous research mainly laid the emphasis on the operating strategy and driving force of social networks as well as the identification of the influential seed nodes,while the control method of information diffusion process,the consumers' behavior choices and the related key influence factors were neglected.Therefore,the values in the diffusion of products' information on social network platforms cannot be fully exploited.This study focuses on the products' information diffusion in the consumers' self-organized social networks,and investigates the consumers' behavior choices,influence factors as well as the diffusion process control of products' information in depth.The primary results and innovative points are listed below.1.This study has conducted a multi-attribute analysis on the information diffusion process.In previous studies,scholars mainly focused on the analysis of the diffusion network structure and took a single influence factor into account.However,information diffusion process is subjected to multiple factors and the complexity is not fully considered.This study investigates information diffusion process from four aspects,namely,network structure,information attribute,users' attributes and the leakage of users' information in transmission process.More specifically,the strong and weak connections in social networks,information overload,controversies and emotions in information,the connection strength among users,the influence of users,the users' transmitting behaviors and the threats to security and privacy are examined in detail;in addition,the effects of the above factors on information diffusion and the influence relations among these factors are also analyzed.This section has provided antecedent analysis for the subsequent modeling of information diffusion process.2.The diffusion model of the products' information is established.This study proposes a multi-point cross communication model based on MapReduce so as to spread the information in a wider range.It should be noted that the cross communication among multiple points is not to transmit information simply and blindly and but to realizes the following two functions:(1)filtering the information in one group;(2)sorting the filtered dataset and finding the most interesting and unique subject for diffusion.On that basis,the Bayesian network(BN)model for describing the diffusion of products' information is established,in which the correlations of influence factors and users' behaviors among the propagation of products' information are analyzed.Further,this study explores the model's three attributes,namely,conditional independence attribute,factorization attribute and the minimum independent set attribute,respectively.The established BN model of products' information diffusion and the related features analysis provide a data structure of knowledges representation and inference bases for subsequent inference.3.This study conducts inference and process control on the diffusion process of products' information.We propose some effective methods for behavioral inference,influence inference and acquiring the effective transmission pathway.Behavioral inference refers to acquiring the posterior probability of the consumers' behavior choice in information transmitting process under a series of influencing factors.This study employes sum-product inference algorithm for generating the information diffusion clique tree,then applies influence passing algorithm in the clique tree and acquires the marginal posterior probabilities of the consumers purchasing,transmitting products' information and taking no actions.Influence inference refers to the process that performing downward information passing operation on the basis of the effects of upward passing algorithm,reversely deriving the marginal posterior probabilities of the influence factors and acquiring the key influence factors.As for the acquisition of effective traces in information forwarding process,the node identification principle for two-dimension code generator and the linked-list implementation method for the correlation between information and nodes are designed;moreover,the safety problem of personal information in information diffusion process is addressed.The products' information designed by two-dimesion code is visible to all consumers during the diffusion,but users' information encrypted to generate the users' identification codes,is invisible to all transferring nodes in the diffusion process expect the two-dimension code generator systems.two-dimension codes of the products' information are designed as public codes while the users' information codes were designed as private codes.Accordingly,not only the threats of personal information leakage in propagation can be solved,but also the diffusion process of products' information can be well controlled.4.For considering the accumulative contributions of the consumers' diffusion behaviors,this study proposes a call queueing method for the consumers' self-organized network based on multi-attribute analysis results of the consumers.This method aimes to award the consumers' diffusion behaviors in social networks and form a virtuous cycle with the consumers' self-organized information diffusion system.The proposed call queening method is designed on the basis of the analysis of the call characteristics in automobile industry,which can shorten the users' call waiting time and provides the consumers in the diffusion network of products' information with high-quality services.Specifically,the multi-attributes analysis of consumers refer to the process of multi-attributes classifying analysis and the users' value contribution to products diffusion according to users' attributes,the the effective traces between the information source and forwarding nodes in the propagation chain,and the acquired posterior probabilities of the consumers' purchasing products and forwarding information by behavioral inference.5.Finally,the mechanism of information identification in two-dimension code generator,the encoding method of both public and encrypted information as well as the linked-list implementation for describing the correlations between information and nodes were designed.The effective transmission pathway in long-chain propagation from value nodes to information source is acquired,and the leakage of users' information in the transferring can be effectively avoided.Moreover,by applying the developed method to the traceable system of defective automobile products,node identifications in the technological process and the linked-list implementation correlating products' information and technological process are designed,so that the defective automobiles can be precisely recalled.