Ir al contenido

Cambios

View changes from to

En el instante 10 de octubre de 2025, 7:19:53 UTC, Gravatar albertogarob:

  • Modificado el título de Graph processing frameworks (anteriormente Graph Processing Frameworks)


  • Actualizada la descripción de Graph processing frameworks de

    This chapter provides a comprehensive overview of graph processing frameworks (GPFs)—software systems designed to efficiently analyze and compute on large-scale graph data. GPFs enable the representation of entities as vertices and their relationships as edges, offering APIs and computation models that abstract low-level parallelization details while supporting diverse applications such as social network analysis, biological systems, and fraud detection. The authors classify more than twenty notable GPFs along four key dimensions: target platform (shared, distributed, or hybrid memory), computation model (MapReduce, BSP, GAS, GSA, etc.), processing approach (vertex-, edge-, subgraph-, or neighborhood-centric), and communication model (message passing, shared memory, or dataflow). Detailed discussions include classical frameworks like Pregel, GraphLab, Giraph, and GraphX, as well as recent innovations leveraging GPU, Hybrid Memory Cube (HMC), and ReRAM architectures (e.g., GraphH, GraphP, GraphR). The chapter also illustrates a practical example of computing the degree distribution of a large social network using the Apache Flink Gelly API, demonstrating scalable parallel execution. In closing, the authors identify open challenges in graph partitioning, hybrid architectures, benchmarking, and performance evaluation, emphasizing that efficient parallelization and resource-aware design remain central to advancing next-generation graph processing systems.
    a
    This chapter provides a comprehensive overview of Graph Processing Frameworks (GPFs), which are software systems designed to efficiently process large-scale graphs across distributed and parallel computing environments. It begins by defining GPFs as toolsets for representing data as vertices and edges, emphasizing their role in managing the three challenges of big data—volume, velocity, and variety. The chapter classifies GPFs along four major dimensions: target platform, computation model, graph processing approach, and communication model. It reviews key frameworks—such as Surfer, Pegasus, GBase, Pregel, GraphX, Giraph, Gelly, GraphLab, GraphChi, and Totem—covering their underlying architectures, data abstractions, and parallelization strategies, including MapReduce, Bulk Synchronous Parallel (BSP), and Gather-Apply-Scatter (GAS) models. A practical example using Apache Flink’s Gelly demonstrates how GPF APIs can be applied to compute the degree distribution of a large social network, highlighting performance scalability with multicore systems. The chapter concludes by discussing future research directions, including graph partitioning in heterogeneous (CPU–GPU) environments, the need for standardized benchmarks, and challenges in performance evaluation and algorithmic efficiency. Overall, it situates GPFs as essential infrastructures for large-scale graph analytics in domains such as social networks, bioinformatics, and cybersecurity.


  • Añadidas las siguientes etiquetas de Graph processing frameworks


  • Modificada la URL de la fuente de Graph processing frameworks de https://doi.org/10.1007/978-3-319-63962-8_283-2 a https://doi.org/10.1007/978-3-319-77525-8_283