Rule Agent-Oriented Scientific Workflow Execution

Over the last decade, scientific workflows have been become a remarkable paradigm, which integrates distributed heterogeneous computational and data resources and assists scientists to perform data management, analysis, simulation in silico experiments. However, compared to traditional business workflows, scientific processes still haven’t been widely addressed and shared because of their extra requirements. In this paper, inspired by the spirit of subject-oriented business process management (S-BPM), we propose a rule agent-oriented approach to model weakly-structured scientific processes, where each agent has an internal behavior and the scientific workflow execution is driven by the messaging between distributed rule agents. As a proof-of-concept implementation we use the Web rule language Prova to declaratively represent the knowledge-intensive scientific logic as semantic rules, wrapped in the agents, and to support message-driven conversation-based interactions between these rule-based agents.