Modeling Service Choreographies with Rule-enhanced Business Processes

Milan Milanoviü
FON-School of Business Administration
University of Belgrade
Belgrade, Serbia
e-mail: milan@milanovic.org
Dragan Gašševiü
School of Computing and Information Systems
Athabasca University
Athabasca, AB, Canada
e-mail: dgasevic@acm.org


Abstract—— The research community has so far mainly focused
on the problem of modeling of service orchestrations in the
domain of service composition, while modeling of service cho-
reographies has attracted less attention. The following chal-
lenges in choreography modeling are tackled in this paper: i)
choreography models are not well-connected with the underly-
ing business vocabulary models. ii) there is limited support for
decoupling parts of business logic from complete choreography
models. This reduces dynamic changes of choreographies; iii)
choreography models contain redundant elements of shared
business logic, which might lead to an inconsistent implementa-
tion and incompatible behavior. Our proposal –– rBPMN –– is an
extension of a business process modeling language with rule
and choreography modeling support. rBPMN is defined by
weaving the metamodels of the Business Process Modeling
Notation and REWERSE Rule Markup Language. To evaluate
our proposal, we use service-interaction patterns and compare
our approach with related solutions.
Keywords - BPMN, business rules, business processes
I. INTRODUCTION
Responding to the increasing demands for developing
advanced solutions to the integration of business processes in
collaborative information systems, service-oriented architec-
tures (SOAs) emerged as a promising approach. Offering
features such as loose-coupling, statelessness, reusability,
and interoperability, the key SOA issues are service publish-
ing, discovery, and composition. Considering the present
state of SOA, we can witness a need for the development of
new software engineering approaches suitable for this new
development context. Here, we consider the challenge of
software engineering languages for service composition.
Following proven principles of business process model-
ing, service engineers have prevalently based their approach-
es on languages such as Business Process Modeling Notation
(BPMN) [19]. Such languages offer a suitable way for re-
quirements elicitation from stakeholders, which can (semi-
)automatically be bound to the existing services and trans-
formed onto the executable service compositions (i.e., lan-
guages such as Business Process Execution Language,
BPEL[18]). In the service composition task, we generally
have two main approaches [30]: i) service orchestration ––
composition of service from the perspective of one of the
participants. Orchestrations are typically modeled w.r.t. con-
trol flows, while workflow patterns
(http://www.workflowpatterns.com) are used as best practic-
es and evaluation framework for comparison of orchestration
languages; ii) service choreographies –– composition of ser-
vices from a global perspective where service interaction is
the primary focus. Similar to workflow patterns, service-
interaction patterns (SIPs) are used as best practices for ser-
vice choreographies and comparison of choreography lan-
guages.
The research community has so far mainly focused on the
problem of modeling of service orchestrations, while cho-
reographies have attracted less attention [17]. In this paper,
we exactly focus on the problem of modeling choreographies
in order to address to the following challenges: i) service
choreography models are not well-connected with the under-
lying vocabulary/domain models. This reduces type safety;
ii) there is limited support for decoupling parts of business
logic (e.g., constraints and process decisions) from complete
choreography models. This reduces dynamic changes of cho-
reography models; iii) service choreography models typical-
ly contain redundant elements of the shared business logic,
which might lead to the inconsistent implementation and
potentially incompatible behavior.
To address the above challenges, in this paper, we pro-
pose a rule-based approach to modeling of service choreo-
graphies. Our solution, rBPMN, is based on the integration
of two proven languages –– BPMN and REWERSE Rule
Markup Language (R2ML) at the level of their metamodels.
In the rest of the paper, we first introduce background know-
ledge, including, business process, service choreographies,
metamodels of BPMN and R2ML, and challenges for inte-
gration of business rules and processes. Next, we introduce
our proposal through a detailed description of the metamodel
(Sect. III), which is followed by the discussion about their
support for SIPs (Sect. IV) and a case study (Sect. V). Before
concluding the paper, we discuss the related work (Sect. VI).
II. BACKGROUND
A. Business processes
According to [30] ““a business process consists of a set
of activities that are performed in coordination in an organi-
zational and technical environment. These activities jointly
realize a business goal. Each business process is enacted by a
single organization, but it may interact with business
processes performed by other organizations.”” BPMN is the
OMG’’s standard for business process modeling [19]. BPMN
has a graphical notation, but has no standardized metamodel.
B. Service Choreographies
According to the Web service glossary, ““a choreography
defines the sequence and conditions under which multiple
cooperating independent agents exchange messages in order
™111000      111444ttthhh      IIIEEEEEEEEE      IIInnnttteeerrrnnnaaatttiiiooonnnaaalll      EEEnnnttteeerrrppprrriiissseee      DDDiiissstttrrriiibbbuuuttteeeddd      OOObbbjjjeeecccttt      CCCooommmpppuuutttiiinnnggg      CCCooonnnfffeeerrreeennnccceee
™444ℤ777777Ω///ℨ      $$$222666...000000      ©©©      222000ℨ      IIIEEEEEEEEE
DDDOOOIII      ℨ...℡000999///EEEDDDOOOCCC...222000ℨ...ℴ
Ω444