Algebraic Optimization for Processing Graph Pattern Queries in the Cloud

MapReduce platforms such as Hadoop are now the de facto standard for large-scale data processing, but they have significant limitations for join-intensive workloads typical in Semantic Web processing. This article overviews an algebraic optimization approach based on a Nested TripleGroup Data Model and Algebra (NTGA) that minimizes overall processing costs by reducing the number of MapReduce cycles. It also presents an approach for integrating NTGA-based processing of graph pattern queries into Apache Pig and compares it to execution plans using relational-style algebra operators.