Workflow considerations in the emerging CI-ML virtual organization

In a virtual organization, the interaction of its members for any purpose generates a sequence of activities referred to as a workflow. This paper seeks to identify the work- flows needed for the Computational Intelligence and Machine Learning Virtual Organization. The underlying architecture of the repository should support these workflows in a smooth and efficient manner. The workflows for the CI-ML VO are categorized as: sharing data, sharing software resources, sharing computa- tional resources, education and networking. These workflow categories, while common for many other VO's, gain some unique qualities when applied to CI-ML. These qualities were considered at the NSF workshop to identify many established and emerging VOs and their characteristics. A common char- acteristic is that each VO has a domain of application, in which most of the members of the VO are familiar. The infrastructure being established by the VO should nurture and sustain the regular and evolving activities of the VO. Therefore, the lifecycle of the many workflows in a VO should be considered in detail. The life cycle of a workflow identifies the resources and interactions between a sequence of logical activities. The computing resources set up for each interaction within the VO is dependent on how these interactions are identified. These interactions are easily defined as a use case to describe how a type of system user is able to interface with the rest of the system. A use case is usually defined with an actor, which can be considered as a user or subsystem, and a definition of acceptable behaviors across system boundaries. The formal definition of all possible use cases for a particular actor is important due to implications on system security and information assurance. The collection of use cases define the functional requirements for the VO and define the services it provides. Given the functional requirements and its accompa- nied workflows a software architecture can be created that is modular and scalable. The importance of creating workflows has major implica- tions for the future growth of a VO. Considerations for the CI- ML VO were made to allow its expansion given the addition of new functional requirements and use cases. This forced the software achitecture to be well defined to accomodate these changes while having minimal impact on the system as a whole. This paper will discuss these workflows, their impact on the software architecture, and security issues to accomodate the use case requirements.