دانلود رایگان مقاله انگلیسی پخش اطلاعات و شایعه - الزویر 2018

INTRODUCTION

Modern societies understand the world, manifest different viewpoints, and test their objectiveness by exchanging information through direct communication or, in more recent years, through online social networks. On a larger scale, this process may also create consensus and mitigate social friction through public debate, two essential aspects of a healthy democracy. Information diffusion is often represented by pieces of information (e.g., news, scientific, or historical facts) that spread through a network. As for the network, that consists of interacting entities such as individuals, institutions (e.g., governments, authorities, or other organizations), and private entities (e.g., media, marketing agencies). The Internet era has offered new means to produce and share information through large-scale online social networks. The disposition of large amounts of data coming from diffusion traces has helped scientific research improve our understanding of diffusion processes arising in various disciplines, including sociology, epidemiology, marketing, and computer system security. However, the democratization of content creation and sharing has not been adequately coupled with effective (self-, collective, or automatic) moderation, correction, and filtering mechanisms. Consequently, the explosive volume of the available content brings forward huge challenges regarding the human capacity to process that fast-paced and gigantic information stream as well as regarding the technical aspects of data management. Our daily information diet tends to promote the variety in the content we consume to the expense of its precision and detail. During moments of crisis, the scarcity of trustworthy information and lack of time to analyze it lead to the proliferation of false rumors. There are also various psychological factors that impact the way we participate in this exchange. For instance, people get influenced by others, but also tend to search and recall information and facts that align with their already formed belief system (confirmation bias). Furthermore, users interact preferably with people of similar profiles and opinions (homophily), a tendency that greatly reduces the heterogeneity of the user’s perceived public debate.

CONCLUSION

The future of the diffusion networks field is full of interesting problems and potential applications. It will continue to enrich our understanding of diffusive phenomena and, at a second level, is expected to also change how information is circulated in online social networks. The subject of this chapter was first to analyze the way information diffusion takes place in modern large-scale online social networks and the challenges regarding the control of certain types of undesired diffusion such as rumors, fake news, and others. We have presented an overview of the complex context in which these information-related diffusive phenomena appear and how individuals participate in the process acting as users of online social platforms. To present the background of related problems, we went through various approaches for modeling information cascades, including the early used virus models and the more recent independent cascades model. Specifically for the latter model, we spoke about its large-scale dynamics and how that relates to the network properties, the existence of a threshold value that defines the point of transition between subcritical and supercritical behavior, and the connection of that threshold value to the spectral radius of the Hazard matrix of the network. Subsequently, we discussed a framework that we proposed recently for spectral activity shaping under the Continuous-Time Independent Cascades Model [13] that allows the administrator for local control actions by allocating targeted resources, which can alter locally the spread of the process. The activity shaping is achieved via the optimization of the spectral radius of the Hazard matrix, which enjoys a simple convex relaxation when used to minimize the influence of the cascade. In addition, by reframing a number of use cases, we explained that the proposed framework is general and includes tasks such as partial quarantine that acts on edges and partial node immunization that acts on nodes. Notably, this generic framework can describe complex strategies that may use several immunization options by deploying simultaneously resources of different types (removal of edges, nodes, partial immunization, etc.). Specifically for the influence minimization that is the one directly related to rumor spreading control, we presented the NetShape method, which was compared favorably to baseline and a state-of-the-art method on real benchmark network datasets. Among the interesting and challenging future work directions, on the same line to the presented framework, there can be the introduction of an “aging” feature to each piece of information that would model its loss of relevance and attraction through time, and the theoretical study and experimental validation of the maximization counterpart of Netshape method.