A Plan Execution System For Web-Based Scientific Data Integration

The sheer quantity of data available to scientists via the Internet is staggering, and more is being collected, aggregated and processed daily. Unfortunately, this data is scattered in collections throughout the network, stored in a variety of formats, and accessible via many different types of programs. NASA is a primary producer of scientific data, as well as an important consumer of such data, and thus developing systems to help scientists access, integrate and process large volumes of heterogeneous data is clearly in NASA's interest. Theseus is a commercial high-performance plan execution system that we have developed for integrating web-based data. We propose to extend the Theseus architecture to create a practical system that will make it easy for earth scientists to access, stage, integrate and process large volumes of scientific data distributed over the Internet. Theseus employs a dataflow architecture for plan execution, which we can leverage to create an innovative, scalable approach for processing distributed scientific data. The resulting system will be used as part of a collaborative ongoing effort with scientists from NASA Ames and University of Montana who are applying AI planning technology to ?Biospheric NowCasting? -- real-time ecosystem modeling from Earth Science data.