Fast datalog evaluation for batch and stream graph processing

Abstract Implementing complex algorithms for big data, artificial intelligence, and graph processing requires enormous effort. Succinct, declarative programs to solve problems that can be efficiently executed batching streaming data are in demand. This paper presents Nexus, a distributed Datalog evaluation system. It evaluates using the semi-naive algorithm batch incremental asynchronous iteration. Furthermore, we evaluate with aggregates determine advantages of implementing iteration on its performance. Our experimental results show Nexus significantly outperforms acyclic dataflow-based systems.