Automaton Meets Query Algebra: Towards A Unified Model for XQuery Evaluation over XML Data Streams

Abstract. In this work, we address the efficient evaluation of XQuery expressions over continuous XML data streams, which is essential for a broad range of applications including monitoring systems and infor-mation dissemination systems. While previous work has shown that au-tomata theory is suited for on-the-fly pattern retrieval over XML data streams, we find that automata-based approaches suffer from being not as flexibly optimizable as algebraic query systems. In fact, they enforce a rigid data-driven paradigm of execution. We thus now propose a unified query model to augment automata-style processing with algebra-based query optimization techniques. The proposed model has been success-fully applied in the Raindrop stream processing system. Our experimen-tal study confirms considerable performance gains with both established optimization techniques and our novel query rewrite rules.