Interactive 3D Tree Visualization Of Network Community Based On Neo4j

Network data is everywhere.However,analysis and understand the complex network relationships is a huge challenge.Thus,it is of great significance to study how to accurately identify the network topology,understand the network function and explore the potential features of the network data.Visualization as a key means of network data analysis,using visual information to enhance the understanding of network structure,as it were,a picture is worth a thousand words.Community structure is the main characteristics of the network,reflects the internal structure and common rules of network.As a result,using the community structure as the main basis the network visualization can reduce the visual complexity and information load.By the constraints of 2D hardware,the mainstream visualization methods were focus on 2D view.With the development and popularization of virtual reality and holographic technology,3D network visualization will become an important way of data expression.In addition,human-computer interaction improved user experience,is an indispensable technical support in data visualization.In this paper,an online interactive visualization tools for detect network community is implemented.Solving the 3D network visualization problem like view overlap and navigation difficulties,revealing the intrinsic structure of the network,has important scientific significance.The main research contents and results are as follows:(1)On the basis of the Neo4j and WebGL,an online 3D interactive visualization platform of network community is designed and implemented.This platform uses Google cache technology to reduce read and update operation on graph database,implements Louvain method to detect community structure,provides interface of data manipulation,community detection and 3D visualization,has good expansibility of clustering algorithm.(2)A new method of 3D network visualization is proposed.The hierarchical community structure was used as the basis for the visual layouts of network,and then mapped it to a 3D botanical tree.Network nodes are mapping to leaves,community are mapping to branches,trees modeling on the process of Louvain clustering algorithm.This method displayed data clustering process explicitly,but also emphasized the overview and detail association of data.(3)Interactive operation in 3D scene,like rotation,zoom,translation and picking,was used to solve view overlap caused by complex branches and leaves in the tree map,reducing visual complexity and eliminating data ambiguity.For the sake of minimizing edge crossing in network,a new way to plan the network relationship is proposed,which the network adjacency between node and its neighbor has to pass theirs Least Common Ancestors.