Model-Driven Development of Relational Development and Semantics-based Scheduling of Concurrently-Executed Workflows

There are two parts of this talk: to present the research problem of my thesis and its proposed solution. The problem being: how to manage the concurrent execution of business workflows via a semantics-based approach? The proposed solution is two-folded and may be presented in a bottom-up manner. The assuming input to the lower level is a set of object models, with each incorporated with the classification, associations, and behaviours (i.e. methods) of business entities. In particular, each object behaviour is specified in terms of a single first-order predicate, encapsulating its exact guard and intention.

At the first stage, the input object (data and behaviour) model is transformed into its relational model counterpart, a ready form targeting for the automatic model-to-text generation into a working SQL database. Focus of this stage, from both the model and meta-model levels, is on the formalisation of the source object and target relational models, as well as the justification on the rules of model transformation.

At the second stage, assuming from the first stage that each business process component (i.e. business activity) is an atomic SQL transaction transformed from its corresponding object method, a framework is developed for deriving the semantics of each individual workflow from those of its component transactions, in order to properly schedule the execution of transactions from a set of concurrently-executed workflows. We shall talk about both intra- and inter-workflow consistency. Intra-workflow consistency concerns about its static launchability, i.e. can the workflow ever be launched? Inter-workflow consistency, in the presence of concurrency, concerns about formally arguing if an interleaving order, composed of transactions across multiple workflows, is both valid and the optimal. Both these two levels of validity checks may be based on the static analysis on the composition of semantics from the set of component transactions.