In today’s digital landscape, complexity grows with increasing data volume and degree of interconnection. A suitable data abstraction is crucial for comprehending and navigating this dense network of connections. Starting from Euler’s pioneering work on The Bridges of Konigsberg in 1735, graphs have steadily evolved as a robust and adaptable conceptual framework. Graphs are universal representations of concepts, where nodes are markers for distinct entities and edges delineate their interrelations, further enriched with detailed annotations when necessary. Graphs are successful in various domains, like bioinformatics, e-commerce, logistics and transportation networks, urban planning, and even pandemic analysis or vaccine development (e.g., during COVID-19).

Although graphs enable complex analysis and decision-making, processing graphs to understand real-world phenomena and to solve real-world problems raises many challenges that threaten to keep graph processing intractable for the current generation of applications. For example, creating graphs from massive data sources or with generative approaches poses multiple challenges, including volume, velocity, and variety. Furthermore, the variability and irregularity of graphs and their processing algorithms challenge the use of established heterogeneous hardware, general-purpose big data solutions, or computing continuum mechanisms. Continuous operation on (streaming) graphs requires new techniques for adaptivity and optimization – e.g., provisioning, allocation, elastic scaling, migration, offloading, partitioning, consolidation, and caching – to be combined across large-scale information and communication technology infrastructure.

Addressing these challenges for graph processing workflows at real-world scale and with societal impact requires a holistic approach, that leverages the expertise and synergies of multiple communities related to graph processing. This Dagstuhl Seminar will provide a unique opportunity for encounters between these distinct communities, each addressing graph processing at scale. We aim to bring together three essential communities: distributed, parallel, and cluster computing, machine learning for, on, and with graphs, and social and information networks. We facilitate therefore a better understanding of each community’s challenges regarding graph processing, and promote a synergic relationship to shape holistic, actionable knowledge of graph processing.

In search of the holistic view of massive-scale graph processing, the seminar prominently features five topics of discussion: (1) massive graph creation with generative and analytical approaches, (2) graph processing algorithms and workflows, (3) graph operations across the digital continuum, (4) adaptivity and optimization to ensure performance, scalability, and sustainability, and (5) applications at real-world scale with near-term societal impact. The seminar will feature keynotes, tutorials, and break-out sessions to facilitate discussions that shape a comprehensive vision and roadmap with regard to guiding research in graph processing for the upcoming years. Concretely, we plan to (i) establish a uniform vocabulary across communities for issues related to graph processing, (ii) identify key open graph-processing challenges and opportunities across communities along with ideas for long-term research, design a holistic approach (blueprint, reference architecture, and experimental methodology) to graph processing in the digital continuum, and (iv) envision flagship applications for holistic graph processing with real-world scale and societal impact. We also plan for long-term, high-visibility collaboration and dissemination, with mechanisms such as roadmap white-papers, networks of excellence, EU-level projects, and follow-up Dagstuhl Seminars.

Creative Commons BY 4.0

Sanja Fidler, Alexandru Iosup, Ana Lucia Varbanescu, and Hannes Voigt