A Comparative Analysis of Large-scale Network Visualization Tools Md Abdul Motaleb Faysal and Shaikh Arifuzzaman Department of Computer Science, University of New Orleans New Orleans, LA 70148, USA. Email: mfaysal@my.uno.edu, smarifuz@uno.edu Abstract—Network (Graph) is a powerful abstraction for representing underlying relations and structures in large complex systems. Network visualization provides a convenient way to ex- plore and study such structures and reveal useful insights. There exist several network visualization tools; however, these vary in terms of scalability, analytics feature, and user-friendliness. Due to the huge growth of social, biological, and other scientific data, the corresponding network data is also large. Visualizing such large network poses another level of difficulty. In this paper, we identify several popular network visualization tools and provide a comparative analysis based on the features and operations these tools support. We demonstrate empirically how those tools scale to large networks. We also provide several case studies of visual analytics on large network data and assess performances of the tools. We show both runtime and memory efficiency of the tools while using layout algorithms and other network analysis methods. Index Terms—Big networks; Visualization; Visual analytics; Network analytics; Graph mining; Scalable algorithms I. INTRODUCTION Network visualization is an important part of graph-based data analysis and research. Networks are often the most convenient way to represent interactions among entities in social, biological, infrastructure and other information systems [7], [8]. Examples include interaction among persons in social networks, connectivity of web pages in world wide web graphs, and interaction of proteins in biological networks [29]. Network analysis and visualization help discover structures and patterns in a network and thereby reveal useful insights [8]. Large-scale network visualization has been a field of re- search interest for more than a few years [3], [1], [5]. There exists a number of tools that offer different functionalities for visualizing, analyzing, and filtering networks [32], [33], [34], [30]. Many of these tools are developed considering a particular application domain such as biological data [1] and social networks [4]. Developer packages are available in programming languages such as python and R to display network data. Some tools provide visualization along with a suite of network algorithms [1], [2]. Some of the tools [30], [31] require programming background while others [37], [39], [42] do not. Some tools provide web-interface [34], [35] while some others [1], [32] run on desktop. Further, the emergence of large network data necessitates the development of scalable tools [5], [6]. The existing tools show a varying degree of scalability. Therefore, there is a need for studying existing network visualization tools based on the general usability, user-friendliness, use case scenarios about how those tools and which of the tools can be used for a specific network data, and scalability for large network analysis. Such study will help a person to decide which tool to use for a specific case based on his need. One such study in [3] compares a couple of tools based on scalability. Comparative studies on other visualization metrics should be conducted to let end users have freedom in choosing the specific tool he needs to use. In this paper, we have chosen five widely used visualization tools for an empirical and comparative analysis. Our comparisons are based on factors such as supported file formats, scalability in visualization and analysis, interacting capability with the drawn network, end user-friendliness (e.g., users with no programming back- ground, cost-effectiveness), and developer friendliness (open- source, multi-platform supports) and support for useful visual analytics. In this paper, we focused mainly on the network visualiza- tion tools that are free, open-source, desktop-based and do not require programming background to perform operations and discover insights from a network. Table I lists the five tools we study along with their versions we used and compatible operating systems. We also provide a description of some other prominent visualization tools in Table II. Next, we present the network visualization tools we chose, discuss the features and components, provide some case stud- ies of visual analytics on real-world scenarios, and conduct comparative analysis on the functional capabilities. II. TOOLS FOR LARGE- SCALE VISUAL ANALYTICS ON NETWORKS We provide a description of five popular network visualiza- tion tools below. A. Cytoscape Cytoscape [1] is an open source software platform for integrating, visualizing, and analyzing network data. The tool originally developed for processing biomolecular interaction networks [16] can also be used for general complex network analysis. Features. Cytoscape is a powerful tool for visualization and analyzing protein-protein interaction (PPI) data in bio- logical network. This tool supports a number of layouts to view the graph data in a convenient way, e.g., hierarchical layout, Circular layout, orthogonal layout, and tree layout. One powerful feature of Cytoscape is its App Store from where a number of applications can be downloaded. Each of those applications comes with different graph analytics and visualization features that serve different purposes such as creating layouts, computing graph metrics, and clustering.