Birds of a Feather Archive

Large-Scale Dynamic Network Analysis: Scalable Parallel Algorithms with Applications

Authors: Sriram Srinivasan (University of Oregon), Boyana Norris (University of Oregon), Sanjukta Bhowmick (University of North Texas), Sajal Das (Missouri University of Science and Technology)

Abstract: We propose to discuss application needs in the area of large-scale dynamic network analysis. Given the latest advancement in exascale computing and the volume of data available, an important problem is how to analyze large-scale networks to be meaningful for applications. Many challenges exist in developing software for analyzing dynamic graphs, including consensus on the output, reproducibility, and most critically whether existing parallel update algorithms support real-world applications’ needs. We aim to take steps toward forming a community of users of dynamic network software while spreading awareness about the tools available and the challenges related to large-scale dynamic graph analysis.

Long Description: Large-scale network analysis has applications in a wide range of domains including epidemiology, biology, cyber security, and, drone delivery system. With the advancements in big data, there has been a plethora of relational data available, i.e., data that can be converted to networks. Analyzing these networks can provide insights into the underlying system. Although the HPC community has developed efficient ways to analyze large-scale networks under different architectures, a critical challenge is to match the available tools and their algorithms to the application needs. Given the exhaustive list of available tools for large-scale networks, there exists a huge gap in understanding of what tools are suitable for which types of networks. Networks have a diverse nature and it is tough to generalize the approach and make that a de-facto standard. This BoF session aims to match the requirements of applications with the tools provided by the state-of-the-art graph analysis software. We will specifically focus on matching algorithm needs to network analysis objectives related to large-scale dynamic graphs. We will invite a panel of 4-5 domain scientists who will highlight the dynamic network analysis requirements of their applications. We have reached out to researchers in epidemiology, cybersecurity, bioinformatics, and smart cities, to potentially speak at the panel. Based on these discussions in the remainder of the panel, we will discuss how well current software addresses these needs, and have an interactive discussion with audience participation on the tools and standards needed for dynamic network analysis software. This session will benefit the community that has access to relational data and HPC resources but lacks the skills to analyze them. This BoF session will outline the challenges, help build a community for dynamic network analysis, and, open up future research directions which can lead to new opportunities for collaboration.

URL:


Back to Birds of a Feather Archive Listing