Research On Nesting Visualization And Analysis Of Graph Data

Graph data exists in modern human society and virtual network society,such as the social network graph data,the scholar cooperation relationship network and the neural connection network.These graph data include a time dimension,which makes this type of data have dynamic characteristics.We call this type of data dynamic graph data or dynamic network data.However,knowledge and information do not have a natural form,which is invisible to the majority of users.The dynamic graph data and the temporal feature information hidden inside need to be processed through the data mining and visual analysis process,then presented to users through a formalization expression.Dynamic network visualization can play a key role in complex network analysis.As the time dimension grows,the amount of dynamic network data will also grow rapidly,meanwhile,the data of dynamic network is constantly updated in the visualization process,and the latest data has an inestimable on the effect of the original visualization.Therefore,the characteristic of the dynamic graph data brings big challenges to the analysis and the visualization.At present,the research of the dynamic graph visualization methods mainly focus on the layout adjustment,the visual element selection and the temporal feature expression.The challenges are mainly focused on:1)The visualization layout of the graph data usually uses the intuitive node-link or matrix layout methods?which are unable to visualize complex information;2)The expression of the time series features mostly uses animation or timeline technology,which is insufficient to express the performance of the overall data characteristics;3)The complex interaction methods increase the cost of learning,and also increase the burden of the information understanding.Therefore,exploring a more efficient and intuitive way to visualize dynamic graph data is an urgent problem,which needs to be solved at present.Based on this background,an alternative type of dynamic graph visualization method is proposed in this paper,which embeds step-by-step information into a hierarchical node-link graph,and gives conversion optimization solutions to the temporal patterns visualization,the comparison layout and the incremental layout,then leverages the convenient interactive operations to enhance the display effect of the information.This paper contains the following four research aspects.(1)Design a visualization method for presenting the temporal features in the dynamic graph data.This paper proposes a multi-dimensional link information representation method,which extracts and generates a new time series link data based on the analysis of the changes in the characteristics of each pair of nodes in the dynamic graph data.This paper analyzes a variety of ways to express the time direction,combines the timing characteristics of the dynamic graph,and uses the radian curve as the way to express the time direction,meanwhile,reflect the relationship between the nodes and the layout of the time series data.This method is straightforward,comprehensible and productive,which can effectively and clearly characterize the timing characteristics of dynamic graph data.We applied this method to real scholar cooperative dynamic graph data,and verified the effectiveness of this method in the task of dynamic graph data time series analysis through user experiments.(2)Design a dynamic graph comparison and layout method.This comparison method extends the nesting dynamic graph visualization method,embeds the dy-namic graph data for comparison into the time sequence unit,and realizes the convenient and intuitive comparison of the dynamic network.On the other hand,a EGIB layout method supports interactions and collision detection is proposed.After the time series networks data is processed by the cluster detection algorithm,the clustering grouping of the network data is obtained,and the clustering grouping result is regarded as a tree structure with only one layer,and the width and height of the visualized screen are combined with the tree structure.Then the Squarified Treemaps algorithm can be used to calculate the segmentation rectangle corre-sponding to each grouping node,and then use the force-directed layout algorithm to layout the nodes in the group.The center of the segmentation rectangle is used as a gravity point in the force-guided layout algorithm.The method can ensure that the position of each cluster group is not affected by the force-directed layout elastic positioning,so that the visualization result does not change the overall lay-out during interactions,thereby improving the immersion of the visual analysis This method is applied to the comparison of NeuroData samples,and received very positive feedback about the convenient comparison from the neurologists.(3)An incremental layout Method for Nesting Temporal network Visualization,which is a dedicated design for adjusting the layout of the nesting visualization while changing the time series data.For the situations,such as the increasing or decreasing in the network membership,the alteration of grouping,the sliding of the interested time window,this method gives specific solutions to each one,which reduced the impact of layout alteration on the visualization effectiveness,while helping user to understand the data more intuitively.This method can maximize the preservation of the image map of the embedded visualized network,and can adaptively adjust the nested timing information.The refinement technology can be applied independently or cooperation with the existing force-oriented layout methods.We applied this method to the whole literature collaborative dynamic graph data and evaluated the results,which show that the method in this paper can efficiently adjust the visualization results when the data changes,and can improve the user's understanding of the data changes.(4)The theoretical method is applied to real data for in-depth analysis.The first data is ADHD brain connectivity data.This paper studies the visualiza-tion methods for the data,including data preprocessing,feature extraction,data contrast visual representation and visual analysis methods.Through the interac-tive system,we analyzed the dataset and found some interesting brain connection model.The second data is the IEEE vis data.Through the visualization results,we deeply analyzed the trend of research hotspots,cooperation relationship.The visualization framework designed in this paper and the proposed layout algorithm visualization algorithm can meet the visualization requirements of most types of dynamic network data.The visualization algorithm in the framework has been optimized according to temporal feature,as far as possible.However,as the amount of network data or temporal information increases,the entire visualization process will cost rapidly growth time,thus,big volume of data samples cannot be processed in the framework.Nesting dynamic network visualization still has many shortcomings,the visualization method proposed in this paper is a useful research in this field.