The University of Southern California Information Sciences Institute's Windward project aims to achieve "Scaleable Knowledge Discovery through Grid Workflows," to allow researchers to analyze data at the same speed it is gathered. Windward will incorporate the ideas of industrial engineering, to bring together machinery and raw materials to create automated workflows. "Significant scientific advances today are achieved through complex distributed scientific computations," explains project leader Yolanda Gil. "These computations, often represented as workflows of executable jobs and their associated dataflow, may be composed of thousands of steps that integrate diverse models and data sources." Gil and her research team plan to achieve Windward's goals by integrating artificial intelligence and gird computing. ISI project leader Paul Cohen has extensive experience in using AI systems for complex data analysis and in developing the Semantic Web, which will be incorporated into the Windward system. Grid computing will allow the construction of the regional, national, or even intercontinental AI structures needed for workflow science. ISI collaborator Ewa Deelman has created a workflow system called Pegasus, which maps large numbers of computations to distributed resources while ensuring optimal performance of the application. AI and grid collaboration has already proved successful in earthquake science at ISI. Researchers will now create new workflow techniques to represent complex algorithms and their differences so they can be autonomously selected and arranged to meet the needs of applications.
