Business Artifacts Research at IBM     


David Boaz photoFenno F. Heath III photoLior Limonad photo photo

Business Artifacts Research at IBM - overview

Business Artifacts (also known as "Business Entities" or "Business Entities with Lifecycles") are based on a novel marriage of data and process, and can be used at different levels of abstraction by a variety of business-level and IT stakeholders, including executives, business managers, business architects, solution designers, IT architects, and systems and software engineers.

Businesses and other organizations increasingly rely on business process management, and in particular the management of electronic workflows underlying business operations. The basic challenge in business process management is to find mechanisms whereby business executives, analysts, and subject matter experts can specify, in an intuitive yet concise way, the framework and specifics of how the operations of a business are to be conducted. The specification should make it easy to develop IT infrastructures to automate the operations as much as possible. It should also permit rich flexibility at two levels. First is the level of individual enactments of the business operations, to permit a high degree of customization of the services supported by the business. Second is the level of business operations models (or schemas), to permit highly flexible evolution of the model, and to permit the easy specification of generic models with numerous specializations (e.g., for different regions or different kinds of customers).

Most business process models are based on activity flows, and it is hard in these models to support the high flexibility needed for managing modern business operations. At IBM Watson and other IBM Research labs we have been working on a fundamentally different paradigm for business process management, which elevates the data managed by the process to the same level of importance as the activity flows. This approach focuses on augmented data records, known as "Business Artifacts", that correspond to key business-relevant objects, their lifecycles, and how/when tasks are invoked on them. In general an artifact begins its life with only a few of its attributes defined, and then more attribute values are filled in (and possibly overwritten) as the BEL instance moves through its lifecycle. An underlying premise is that at any point in the processing, all business-relevant information about a artifact instance should be stored in that instance; this might be achieved for example by using a state-machine based specification of how tasks are applied to artifact instances, and explicitly storing the current state in the artifact instance. This premise implies that relevant information about an artifact instance is never "hidden" in the instance's current position in an activity flow.

The business artifact approach provides a simple and robust structure for managing business operations, and has been demonstrated in practice to enable significant efficiencies in business transformation (that is, when performing an overhaul of the organization of the operations of some or all of a business). More generally, the artifact approach substantially enriches communication between business operations stake-holders as compared with the communication enabled by traditional activity-flow based approaches. This is because the artifact approach allows for the explicit specification of business-relevant semantic information (about artifact data types and their lifecycles) which cannot be specified easily in typical activity-flow based models. Based on work that began in 2003 and earlier, the team at Watson has already developed a substantial framework around the artifact paradigm using finite-state machine based lifecycles, including a toolkit for implementing artifact-centric business operations models that leverages existing IBM software products; a method for discovering, specifying, and implementing artifact-centric business operations models; and Siena, a light-weight prototype system for rapid design and deployment of BEL business operations models.

More recently, the Watson team, in collaboration with the IBM Haifa Research Lab, has been developing a new, declarative meta-model for specifying artifact lifecycles, called Guard-Stage-Milestone (GSM). The approach is based on the constructs "guard", "stage", and "milestone". It supports hierarchy, and supports the spectrum from highly prescriptive (a la BPMN) to largely human-driven (as in case management). The hierarchical structure of the stages enables a modular organization of the ECA-like rules that control when stages open, when stages complete, and when milestones become achieved and/or invalidated. The publications listed above present initial work on the meta-model, including mathematical foundations for the operational semantics of GSM. The team is currently developing the Barcelona prototype engine, that supports the GSM meta-model, creating pilot applications using Barcelona, and extending the meta-model to incorporate constructs (e.g., flowchart arrows) to make business process specification more intuitive. GSM and Barcelona are being used by the EU-funded Artifact-Centric Service Interoperation (ACSI) project.

Note: This page is under construction.