数据─网络：场建模与仿真

A data field model of dynamic network is built from the point of view of physics，in which precisely describes the evolution progress of complex dynamic network. Furthermore, the simulative evaluation theory and methods are provided. After building this model, algorithms of several data mining tasks on dynamic networks are designed and the experiment results are analyzed. Cluster of data's attractive and repulsive forces are used to express dynamic networks, and knowledge discovery and data mining algorithms are designed based on them. Methods of dynamic network published are some extensions of research thoughts in static network, in which the time character of dynamic network are not considered. The purpose of our project is to build a data field model of dynamic networks and calculate virtual attractive and repulsive forces between vertices in networks based on considering dynamic networks as a three dimension Euclidean space and adopting time character of dynamic networks through a series of theories, methods and skills. Besides, the displacement caused by them are also calculated. Our time-varying data field model will focus on both the relationship and interaction between nodes and also the networks' changing tracks and progress at different moment. This model can be used to deal with some hot research issues in dynamic pattern field, which would put forward related solving methods from a new point of view and is expected to acquire good results. Our research includes: the methods and theories to adopt attractive and repulsive forces in dynamic networks, accurate mathematical expression and nuclear clustering algorithm design of time-varying data field model, reliability evaluation of simulation experiments based on the data field model of dynamic network and model optimization.

从物理学角度出发，精确刻画复杂动态网络的变化过程，构建动态网络的时变数据场模型(DFDN)并给出仿真评估理论和方法,数据引斥力场聚类表达动态网络并以此设计知识发现与数据挖掘算法。 传统的动态网络研究方法都是基于静态网络相关研究思想的拓展，未考虑动态网络的时序特点。而本课题的目的是，将动态网络看作是三维欧式空间下的力学系统，经一系列理论、方法、技术步骤引入动态网络的时间特性，通过计算网络节点间的虚拟引斥力和由虚拟力导致的节点位移变化，从而构建出动态网络的数据场模型。该时变数据场模型既关注节点个体之间的连接关系和相互影响，又强调不同时刻网络随时间变化的轨迹及变化过程。该模型有望处理动态模式挖掘领域的若干热点研究问题。 研究包括：动态网络中引入引斥力关系的理论与方法、严格的时变数据场模型数学表达及核聚类算法设计、基于动态网络时变数据场模型的仿真实验、可信评估及模型优化等内容。

从物理学角度出发，刻画复杂动态网络的变化过程，构建动态网络的数据场模型(DFDN)并给出仿真评估理论和方法。.传统的动态网络研究方法都是基于静态网络相关研究思想的拓展，未考虑动态网络的时序特点。而本课题的目的是，将动态网络看作是三维欧式空间下的力学系统，经一系列理论、方法、技术步骤引入动态网络的时间特性，通过计算网络节点间的虚拟引斥力和由虚拟力导致的节点位移变化，从而构建出动态网络的数据场模型。该数据场模型既关注节点个体之间的连接关系和相互影响，并强调不同时刻网络随时间变化的轨迹及变化过程。.研究包括：复杂网络数据核力场时间序列特性研究；复杂网络引力场建模技术及其应用研究；结合网络引力场模型和社团结构的检测方法研究；引力场节点团吸引密度的链接预测研究；复杂网络上的链路预测及不对称性演化研究；复杂网络场信息传播模型及其应用研究；复杂动态网络中引入引斥力关系的理论与方法；节点间引力的复杂网络重要节点发现算法研究；基于数据场的复杂网络节点影响力建模、仿真与聚类应用分析；数据场模型数学表达及算法设计、基于动态网络数据场模型的仿真实验可信评估及模型优化等内容。