Edge Bundling Algorithm For Multilayer Node Connection Graphs

Node-link graphs often map entity objects in data to nodes and the relationships between them as edges.Node-link graphs can clearly display complex connections between entities and are therefore widely used in fields such as network topology analysis,social network analysis,financial transaction analysis,and biological structure analysis.The significant hidden value behind them has garnered attention from both academia and industry.In recent years,due to the increasing scale of data,directly visualizing node-link graphs results in a large number of crossed and overlapping edges,causing serious visual clutter for users.The visual clutter in node-link graphs not only affects the aesthetics of the visualization,but also makes it more difficult for users to understand and analyze the data.To address this problem,researchers have proposed edge bundling algorithms that bundle adjacent edges to reduce crossings.However,current edge bundling algorithms ignore the connections between important nodes in the graph,and bundling results can easily lead users to perceive incorrect data.In addition,when there are too many nodes in a limited space,the bundled edges can cover a large number of nodes in the graph,making it more difficult for users to observe the connections of obscured nodes.At this point,the advantages of the edge bundling algorithm are difficult to demonstrate.To solve these problems,this paper creatively combines edge bundling algorithms with cohesive subgraph mining technology to propose edge bundling for multi-layer node-link graphs.The algorithm first extracts subgraphs with different core values in large-scale nodelink graph using subgraph mining technology,then divide the subgraphs into different layers of 3D space,and finally perform edge bundling on the edges within the subgraphs and the edges connecting different subgraphs separately.To clearly display the internal structure of subgraphs and the connections between different subgraphs,two different edge bundling algorithms were designed in this paper:one for 2D space based on betweenness and homophily strategies,and one for 3D space based on adaptive force.To demonstrate the effectiveness of the algorithm,the proposed algorithm was quantitatively evaluated on three public datasets compared to current advanced edge bundling algorithms,and the results showed that the proposed algorithm not only significantly reduces point(or edge)crossings but also fully reflects the connection trend of short edges in the graph,and clearly displays the important subgraph structures and connections between different subgraphs.Based on the problem that the current edge bundling algorithm lacks effective interaction,which makes it difficult for users to interactively analyze the local details of the node-link graph,this paper designs an interactive graph analysis system based on the edge bundling algorithm.The system can deeply explore and analyze node-link graphs through a series of interactive methods such as associative interaction,direct interaction,and context interaction.To verify the usability of the system,a case study was conducted based on a real dataset,and the results showed that the system can help users observe advanced patterns in the graph clearly and provide details of local areas in the graph.