Parallel Processing of Temporal Anti-Joins in Memory

Efficient and scalable processing of temporal anti-joins remains a significant research challenge in temporal databases. To address this issue, this paper introduces a novel temporal primitive designed for transforming a temporal anti-join, including conjunctive equality predicates on non-temporal attributes, into an equivalent algebraic expression involving a temporal inner join. The rationale behind this transformation is that the new expression can be decomposed into subtasks, allowing for parallel execution across multiple CPUs. Experimental results using real-world datasets demonstrate the superior efficiency and scalability of our solution for in-memory processing compared to existing solutions.