Supporting property graphs in apache giraph

摘要

Apache Giraph is a powerful tool for processing very large graphs in distributed environments. One of its main features is a novel programming model that facilitates the design and execution of graph-oriented algorithms in distributed environments. Despite its benefits, using Giraph is a complex and challenging process. Moreover, the Giraph data model (based on labeled directed graphs) makes its use difficult in application domains where nodes and edges contain metadata. This article presents an extension of the Giraph API to provide support for Property Graphs, that is, graphs whose nodes and edges could have properties. In particular, we define a formal method to transform a Property Graph into a Giraph graph, we describe an API for manipulating and querying Property Graphs in Giraph, and we present experimental results that show the applicability and efficiency of our extension.