Stateful Publish-Subscribe for XML Data Streams

摘要

With funding from this grant, we obtained deep understanding of stateful publish-subscribe technology with a special emphasis on data in XML format. We made the following four discoveries: (1) Time is crucial when talking about state in a system. We developed a full characterization and axiomatization of the temporal model in event stream processing languages. (2) We developed a novel language with formal semantics for stateful publish- subscribe. (3) We developed novel algorithms for stateful XML publish-subscribe that achieve two orders of magnitude performance improvements over previous work. (4) We developed a novel framework for the joint optimization of the processing of many stateful publish-subscribe queries. Technology developed with this project has been transitioned to Microsoft where commercial product is being developed.