| Large-scale scientific data management and analysis usually relies on many distributed scientists with diverse expertise. In recent years, such a collaborative effort is often composed and automated into a dataflow-oriented process, a so-called scientific workflow. However, existing scientific workflow tools are single-user-oriented and do not support collaborative scientific workflow composition, execution, and management among multiple distributed scientists. In this thesis, we report our study of collaboration protocols towards building a tool supporting collaborative scientific workflow composition. Based on a scientific collaboration ontology, we propose a service-oriented collaboration model supported by a set of collaboration primitives and patterns. The collaboration protocols are then applied to support effective concurrency control in the process of collaborative workflow composition. We also report the design and development of a service-oriented collaborative scientific workflow composition tool based on an open-source, single-user tool Taverna. |
