ME training

1
Metamodeling and method engineering Metamodeling
languages

• Juha-Pekka Tolvanen
• 3.3.2004
• Lecture 4 Metamodeling languages
• Contents
• Definitions
• Examples of metamodeling languages

2
Defining metamodeling and metamodels

• Models play a crucial role in ME, as in all
  engineering.
• A metamodel is a conceptual model of an ISD
  method.
• one level of abstraction higher than normal
  models
• Metamodeling is a modeling process which takes
  place one level of abstraction and logic higher

3
4 level architecture of IRDS by ISO

• IRDS Reference model

• The IRD Definition Schema Level
• A metametamodel according to which the IRD
  Definition level objects can be described.
• e.g. concept (Wijers 1991) or OPRRs Object
  metatype (Smolander 1992).
• IRD Definition level
• Schemata and application programs specs
• Metaclass level of languages such as Smalltalk.
• Metamodels
• e.g. specification of UML
• The IRD level
• DB schemata and app. programs
• Class level of class-based languages
• Models
• e.g. Customer entity
• The application level
• Application data and execution.
• Instances of class-based languages.
• Instances
• e.g. Customer Juha-Pekka

4
Model architecture (OMG)
Metametamodel
Relationship
Property
Graph
Object
Role
Metamodel
Button
Alarm
InstanceOf
InstanceOf
InstanceOf
Model
Wake
Down
Sets
InstanceOf
System in operation
Down button pressed
5
ISOs IRDS framework

• Information Resources Dictionary Standard shows
  that
• at least four levels of instantiation are
  necessary
• to integrate the modeling of the usage and
  evolution of Iss
• with interlocking pairs (instance-type pair)
• Method Engineering focus on
• defining IRD Definition level (i.e. metamodels)
  based on
• IRD Definition Schema level

ME
ISD
Use of IS
6
An instance example starting point for
metamodeling

• Instance (model)
• What is the structure of the method for modeling
• activities/tasks,
• order between tasks,
• material flow,
• information flow,
• rules for flows,
• triggers for initiating tasks,
• (references between models)

7
An example metamodel

• Types (metamodel)
• Includes all the type information
• object (types)
• relationship (types)
• property (types)
• ...

8
Type-instance pair Some mappings between the
levels
9
An example a modeling method based on the
metamodel
10
Use of metamodels

• Metamodels make methods explicit!
• we can discuss and analyze them
• Systematize and formalize method knowledge
• CASE tool and repository implementation
• mapping to generators and other tools
• interfacing tools (e.g. CDIF XMI)
• compare methods
• support standardization efforts (e.g. OMGs UML)
• Types of metamodels
• Meta-data models (static structure of method)
• Process models (dynamic part of method)
• No modeling is possible without some sort of
  (explicit or implicit) metamodel!

11
First metamodeling task DFD

• What (meta)elements you can detect from the DFD
• Objects?
• Relationships?
• Rules?
• Linkages to other models?
• Structure Chart?
• Entity-Relationship Diagram?
• Others?
• Mappings to
• code?
• testing material?
• ?

12
Implementing metamodels

• Metamodels support implementation of
  situation-specific development tools (modeling,
  interfaces, transformations)
• Metamodel of ERD ERD Modeling (based on
  the metamodel)

13
Example metamodel Booch
14
Booch in use
15
Metamodeling languages

• Metamodeling is based on languages too!
• These vary from purpose
• illustrating vs. formalizing methods
• build tool support?
• meta-data model v.s. process model vs. etc.
• What kind of representation for metamodels
• graphical (ER, NIAM, OPRR, GOPRR, MOF)
• matrix (O/A Matrix),
• text (ObjectZ, MDL, MEL, MOF), or
• form based (GOPRR)

Discussion Which of these should be preferred?
16
A short review to modeling power of metamodeling
languages

• Example from object-oriented design method
• the life-cycle of class instances must be
  specified with one or more state models.
• A state model contains states and transitions
  between two states.
• A state must be specified by a name and a class
  may have only one state with a given name.
• Each transition must be specified with an action
  which is executed when a transition occurs.
• An action is specified as an operation of a class.

17
Partial example STD
18
Metamodeling languages ER

• What is missing from the metamodel
• Different modeling languages? (mapping of action)
• Transitions should be always connected to states?
• Mandatory state name?
• Unique names?
• ?

19
Metamodeling languages OPRR

• What is missing from the metamodel
• Different modeling languages?
• mapping of action to an operation of a class
• mapping of state (model) to class
• Mandatory state names?
• Unique names?
• ?

20
Metamodeling languages CoCoA

• What is missing from the metamodel
• Transition and state mapping to (notation)
  different types
• Dependency of operations to actions
• Mandatory attributes
• Unique states
• ?

21
Metamodeling languages NIAM

• What is missing from the metamodel
• Transition and state mapping to (notation)
  different types

• Interconnected methods?
• Linkage between action and operation
• ?

22
Metamodeling languages GOPRR

• What is missing from the metamodel
• How many state models a class can have?
• Uniqueness of states (for one class)?
• Mandatory values?
• ?

23
Use of meta-data models in ISD

• By specifying the conceptual structure of a
  method metamodels can cover support for
• Abstraction and modeling
• modeling editors and other tools (reuse, naming
  conventions)
• Consistency checking and maintenance
• change of a class name in one place reflects
  elsewhere, etc.
• integration and linkages between methods
• Form conversion
• code generation, testing material, etc.
• exports to other tools
• Review
• documentation, summary reports

24
Metamodeling approaches

• Metamodels address partial aspects of the method
• few values assumptions
• some decision making
• history
• few participation team
• communication, roles
• quite a many processes
• quite many notations, and
• mostly conceptual structures
• We focus to the last two!

25
Meta-data modeling

• A metadatamodel specifies
• languages and representations,
• links between the language constructs,
• consistency checks, and instance population rules
• transformation rules between the language
  constructs (syntax, representation style)

26
Process (or meta-activity) modeling

• A process model specifies
• development tasks, e.g.
• phases
• tasks
• steps
• reviews
• milestones
• checking
• transformations
• decisions
• sequence and composition of tasks
• conditions and rules between tasks

27
Agent models

• An agent model specifies
• usage of CASE environment, human and technical
  agents
• organization structure role models
• communication model