Meta-model

Another approach is to develop a global model that contains plausible models as special cases, defined by alternative values of particular parameters. This converts model uncertainty into uncertainty about the model parameters. Again this can be done using either Bayesian or non-Bayesian approaches. This approach is favored by Morgan and Henrion (1990), who describe how it can be applied to uncertainty about dose-response functions (threshold versus nonthreshold, linear versus exponential). 

---

The approach taken is loosely based on the input-process-output meta-model utilized to transform a problem statement into a functional process. The section Scope definition discusses the intended purpose and potential constraints of the isolation effort, followed by an overview of the Toolbox available to the practitioner (input). The section Method development scouting and scale-up reviews platform-based, highly automated approaches to selectivity scouting, development of the isolation as well as options for scaling up the chromatographic separation depending on purpose and constraints (process). The final section. Performing the task, explores a work breakdown structure approach to the preparative isolation of impurities as a unit operation in the development process (output). 

The approach described reflects an industrial view of the preparative isolation of impurities based on a meta-model developed with the aim of implementing a functional process while maximizing success under a given set of constraints and mitigating risk. This chapter is a snapshot of its current state of evolution. 

METABOLIC CONTROL ANALYSIS METABOLONS META-MODEL FMN,... 

OXYGEN, OXIDES 0X0 ANIONS META-MODEL METAPHOSPHATE ACYL PHOSPHATES OXYGEN, OXIDES 0X0 ANIONS METASTABLE STATE JABLONSKI DIAGRAM TRANSIENT CHEMICAL SPECIES Metastable triplet state,... 

Meta-models a global model is developed that contains plausible models as special cases, converting model uncertainty into uncertainty about model parameters. Again, this can be done using either Bayesian and non-Bayesian approaches. 

Table 3. The important factors found from sequential bifurcation under two meta-models for the Old supply chain simulation. Details of the factors are given in Table 2...

Placed in a completely different and far more fixedly structured domain, the results of the DWQ project were not directly transferable to the CRC 476 IMPROVE. For the support of creative design processes, different approaches were - and are - necessary. Yet the approach of applying methods of meta modeling in domain and application models was expected to succeed there as well. Especially the use of ConceptBase, as described before, was used to achieve a sound conceptual and ontological basis for the modeling aspects, the process extensions to Data Warehousing as researched in the subproject Cl, and the meta process and product repository of the subproject B1. 

An outstanding property of O-Telos is its capability for meta modeling The language supports not only the definition of classes but also the definition of rneta classes (and even meta meta classes)] this leads to a layered model structure where the entities on each level are instances of the entities on the level above. Rules and logical constraints can be defined at any abstraction level. Constraints specify conditions on the specialization and instantiation of classes. Rules make implicit information explicit and thus derive new information from the asserted facts. Within each level, binary relations between the model entities can be indicated through so-called links, and the entities can be further characterized by attributes. 

The remainder of Subsect. 2.6.2 is organized as follows firstly, the general dependencies between the individual submodels are described next, the interrelations between the submodels top-level concepts are clarified and finally, the function of the superordinate Meta Model is discussed. 

The Meta Model is defined on top of the CPEDM. Like the submodels of the C EDM, it is implemented in OWL to enable a formal model integration. 

The Meta Model is imported by all four submodels, as indicated in Fig. 2.36. That way, the ontological assertions of the Meta Model are included in the submodels. 

Model. They form the topmost layer of the concept hierarchy all other classes and relations - those in the Meta Model as well as those in the submodels - are specializations (i.e., subclasses or subrelations) of these fundamental concepts . RelationClass is a good example of such a fundamental concept The class is a means for representing n-ary relations in OWL (by default, the OWL language provides language primitives for binary relations only). Specializations of RelationClass are utilized in all four submodels of the C EDM. 

While the Meta Model is highly useful during model design, it is less relevant for practical applications. Under certain conditions, the Meta Model may even prove harmful, as its highly abstract concepts may confuse an inexperienced user rather than support him/her. For that reason, the interconnectivity between the Meta Model and the C EDM needs to be minimized such that they can be separated easily, once the design has been completed. To this end, the classes and relations defined in Meta Model are not directly used within the individual submodels of the C EDM. Rather, they are redefined and subsequently linked to the original concepts in the Meta Model. Thus, only the links to the Meta Model need to be disconnected if a stand-alone usage of the C EDM is desired. 

In the AToM project [627], modeling tools are generated from descriptions of their meta models. Transformations between different formalisms can be defined using graph grammars. The transformations do not work incrementally but support user interaction. Unlike our approach, control of the transformation is contained in the user-defined graph grammars. 

Domain-specific knowledge is formalized by a process model definition (cf. Sect. 2.4) which constrains the process model instances to be maintained at project runtime. As a consequence, the manager may compose task nets from predefined types and relationships. The process model definition is represented in the Unified Modeling Language UML [560]), a wide-spread standard notation for object-oriented modeling. A process model is defined on the type level by a class diagram which has been adapted to the underlying process meta model for dynamic task nets [388, 389]. 

PROGRES as well as its modeling environment, which offers a graphical editor, an analyzer, an interpreter and a code generator [414], Both the process meta model and process model definitions are specified in PROGRES. The former was created once by the tool builders of AHEAD the latter ones are generated automatically by the modeling environment (cf. Subsect. 3.4.3). 

The process meta model introduces the language (or meta schema) for process model definitions. The meta model is based on djmamic task nets. It provides meta elements for structural (tasks, control and data flows etc.) and for behavioral aspects (e.g. state machines for tasks) of these task nets. 

On the untyped level (Fig. 3.75a), there is no constraining process knowledge (ad-hoc process). The process manager may create and connect any number of tasks with the help of unconstrained types (which are constrained only by the underlying meta model). 

---