An agent-based approach for capturing and linking provenance in geoscience workflows

• We address limitations in e-Science provenance capture.
• We extend the usage of an existing service provenance ontology.
• Our approach is grounded in Semantic Web standards and W3C provenance ontology.
• Our evaluation identifies an efficient solution capable of scaling to large number of workflows.
• We explore the integration of Semantic Web and multi-agent systems to address outstanding challenges in e-Science.

Provenance is becoming increasingly important as web services and computational workflows enable new methods by which work is conducted. Yet, there exist sets of questions that cannot be addressed by current provenance capture systems. We address these challenges by leveraging a service provenance ontology that captures execution details of workflow constituent web services. The ontology is used in conjunction with a multi-agent system to automate provenance aggregation and collation. The use of a multi-agent system eliminates the need to modify service interfaces, as was done in previous research. Simulation experiments are used to evaluate multiple agent topologies and identify an efficient and scalable system that scales to large numbers of workflows.