Research On Ranking Algorithms Of Nodes In Time Evolving Networks And Its Applications

Over the past two decades,network science has developed rapidly and emerged as a multideciplinary research filed.Because of the ubiquity of complex systems in nature and society,complex network,as a simple yet powerful tool to represent and analyze a wide range of complex systems,has been studied in various scientific and engineering problems.The problem of ranking the nodes in complex network is crucial for many realworld problems,and has significant research and application value,such as identification of influential spreaders in networks,ranking of web pages in web search engines,and ranking of products in online socioeconomic systems.The ranking of nodes in complex networks has been widely studied,a variety of ranking algorithms have been proposed under diverse scenarios and for different goals.To date,most of the well-established node ranking algorithms assume that the network topology is static and does not evolve in time,as a consequence,they exhibit important shortcomings when applied to real networks that rapidly evolve in time.The understanding of algorithms' bias introduced by network evolution has practical significance and deserves explorations.In the meantime,previous studies showed that ranking algorithms have potential impact on further evolution of systems,such as search engines.While many ranking algorithms have been extensively studied,we still lack a clear understanding of their potential impact on future evolution of systems.In this thesis,firstly the node ranking algorithms in evolving networks were introduced.Then we explored the long-term impact of ranking algorithms on evolving networks,and provided an application of ranking of nodes in a real-world system.The main contributions of this thesis are as follows:1.We reviewed the relevant basic concepts in network science,several wellknown ranking algorithms in static network,and the potential problem of the ranking algorithms when applied to evolving networks.We then introduced an effective technology to modify the temporal bias of static ranking algorithms.2.We studied the long-term impact of ranking algorithms in evolving networks.A new model of network growth was proposed,based on the proposed model,we introduced a new method to study the long-term impact of ranking algorithms in evolving networks.A static ranking algorithm was compared with its temporal version.This work formed a new framework to evaluate ranking algorithms of nodes.3.By analyzing review datasets,a case study of application of node ranking in online socioeconomic systems was illustrated.This part innovatively proposed an algorithm to distinguish user preference.We found two new kinds of users with difference behavior characteristics which respectively refered to as Rising Trend Anticipators and Declining Trend Anticipators.