The Bio-jETI Framework - User-Level Workflow Design: A Bioinformatics Perspective - page 49

Synthesis Algorithm
The synthesis algorithm then interprets the SLTL formula that specifies the
synthesis problem over paths of the synthesis universe, that is, it searches the
synthesis universe for paths that satisfy the given formula. The algorithm is
based on a parallel evaluation of the synthesis universe and the formula (for
details on the algorithm, cf. [305]). It computes all service compositions that
satisfy the given specification.
2.3.2 Domain Modeling
During the domain modeling phase, as shown Figure 2.9, PROPHETS is pre-
pared for the application domain. Roughly speaking, the domain modeling
involves everything that is required prior to domain-specific workflow devel-
opment, such as service integration and providing meta-information about
the services and data types of the application domain. This activity requires
an overall understanding of the specific domain, as well as explicit knowledge
of the available resources, and in particular concerning their interplay. Thus,
the task of a domain modeler, whose role it is to formalize this knowledge in
terms of ontologies and other kinds of formalisms, is different from the task of
the workflow developer, who exploits the domain model when working with
the readily customized workflow framework.
More precisely, setting up a domain for PROPHETS involves four major
steps (as depicted on the left hand side of Figure 2.9):
1. Integration of services.
2. Definition of the available services and data types, and hierarchical struc-
turing of the domain by semantic classification of types and services via
taxonomies.
3. Interface description of services in terms of their input/output behavior.
4. Formulation of domain-specific constraints, such as ordering constraints,
general compatibility information or dependencies between services.
In the jABC context, service integration simply means the provisioning of
SIB libraries as explained earlier (cf. Section 2.1). The other three aspects of
domain modeling are detailed in the following.
Service and Type Taxonomies
PROPHETS supports the definition of type and service taxonomies as used
by the synthesis algorithm. Technically, the taxonomies are stored in OWL
format [276], where the concept Thing denotes the most general type or ser-
vice and the OWL classes represent abstract classifications. The concrete
types and services of the domain are then represented as individuals that
are related to one or more of those classifications by instance-of relations.