Combining Flexibility and Data Handling in Business Process Models

Flexibility and data handling are two hot topics in current research on business process models. However, these issues are tackled separately through declarative languages and artifact-centric approaches, respectively. It follows that the situations in which the choice of the task to perform is a human decision affected by the presence of suitable input entities are not adequately handled; unfortunately, these situations are common in several industrial processes such as build-to-order ones. An integrated approach able to combine flexibility and data handling is needed and to this end this paper presents a notation named ENTA (ENtities and TAsks): it makes both business entities and tasks first-class citizens in process models and it provides high-level descriptions of the tasks in terms of intended effects and constraints to be met. Some examples of order handling processes are used to explain the issues of entity selection and task selection: deciding which customer orders can be associated with the same supplier order is a case of entity selection while deciding whether to generate a new supplier order for the customer orders selected or to add them to an existing supplier order is a case of task selection.

VERIFICATION OF CONJUNCTIVE ARTIFACT-CENTRIC SERVICES

An artifact-centric service is a stateful service that holistically represents both the data and the process in terms of a (dynamic) artifact. An artifact is constituted by a data component, holding all the data of interest for the service, and a lifecycle, which specifies the process that the service enacts. In this paper, we study artifact-centric services whose data component is a full-fledged relational database, queried through (first-order) conjunctive queries, and the lifecycle component is specified as sets of condition-action rules, where actions are tasks invocations, again based on conjunctive queries. Notably, the database can evolve in an unbounded way due to new values (unknown at verification time) inserted by tasks. The main result of the paper is that verification in this setting is decidable under a reasonable restriction on the form of tasks, called weak acyclicity, which we borrow from the recent literature on data exchange. In particular, we develop a sound, complete and terminating verification procedure for sophisticated temporal properties expressed in a first-order variant of μ-calculus.