Meta-modeling: concepts, tools and applications

Meta-modeling: concepts, tools
and applications
https://dl.dropboxusercontent.com/u/6656530/tutorial_rcis2015.html
(1) Associate Professor
Telecom Ecole de Management, Department of Information Systems
Evry, France
(2) Invited Researcher
Centre de Recherche en Informatique, University of Paris 1 La Sorbonne
France
Saïd Assar 1,2

1. Introduction & background (cont'd)
2
Telecom Sud Paris
2
• INSTITUT MINES - TELECOM: 4 leading schools in
engineering and management
Telecom Paris
Telecom Bretagne
Telecom Ecole de
Management

1. Introduction & background (cont'd)
33
 Since 1999, long-term research collaboration with CRI at
Paris 1 Sorbonne

Learning objectives
 To understand what is a meta-model, its historical
background, when it is needed and how it can be defined
and exploited.
 To witness and participate in building concrete meta-
models and exploiting them using existing meta-modeling
technical infrastructures
 To be aware of problems and challenges in meta-
modeling and the actual state-of-the-art regarding these
issues.
4

Agenda
5
1. Introductory examples
2. Abstraction levels and the instantiation problem
3. Computerized tools for meta-modeling
4. Illustration
a) A Domain Specific Language (DSL) for IoT v3.0 Framework
b) A DSL to model software development processes
c) A DSL to specify web applications
5. Research challenges
6. Conclusion
5

1. Introductory examples – what is a model ?
6
"A model (M) for a system (S) and an observer (O) is any kind of representation
which can help O in answering questions and reasoning about S“
 It is an abstraction, a reduction of reality to its most relevant aspects
 It is the result (i.e. the product) of some specific process
 It is built using some kind of notation (formal and/or graphical)
 Descriptive vs. prescriptive models
 Models are artifacts, i.e. they can also be modeled
Observer

1. Introductory examples – what is a meta-model ?
7
Source: Hanns-Alexander Dietrich, Exercice 1 Meta-modelling, European Research
Center for Information Systems, Information Modelling Summer Term 2012.
Let’s consider this very simple process model with two concepts (event,
function) and one link among concepts (event – function)

1. Introductory examples – what is a meta-model ?
8
Source: Hanns-Alexander Dietrich, Exercice 1 Meta-modelling, European Research
Center for Information Systems, Information Modelling Summer Term 2012.
A meta-model would be a textual,
graphical, and/or formal representation of
the concepts and how they are linked
InstanceOf
Questions
 How to represent a meta-model?
 How can a MM be leveraged in IS
engineering?

1. Introductory examples – generic models (1)
GR model
GB model
RUS model
IND model
IRAN model
Québec model
New York model
TR model
Suppose a trans-national car tracking
system
 interoperability issue
 system evolution issue
Car
plate
model
IsA
9
 Are all these plates special cases
of a car plate model?

1. Introductory examples – generic models
GR model
GB model
RUS model
IND model
IRAN model
Québec model
New York model
TR modelVehicle
identification
model
InstanceOf
 Or is there a generic vehicle ID
model from which these models -
and may be others - can be
derived?
Interactive question:
In terms of system evolution,
these two modeling solutions are
similar or very different?
10

1. Introductory examples – generic models
11
In terms of system evolution, these two modeling solutions are similar or very different?
InstanceOf
IsA
InstanceOf
InstanceOfWhat about the country code,
how to model it?

1. Introductory examples – generic models (2)
12
Method M1 for simple projects
Method M2 for
large projects
Agile method M3 for
risky projects
S1
S2
S3
SN
…
L1
L2
L3
LM
…
G1
G2
G3
GK
…
General
method
IsA ?
General method
structure
InstanceOf ?
For a software development company
with different methods
 evolution issue
 enactment issue

1. Introductory examples – IoT framework
In the field of Internet of Things (IoT), a large set of heterogeneous objects are
expected to interact and be part of a huge information network …
 Is it possible to have a general model for all interacting objects?
13

1. Introductory examples – IoT framework
Source
http://www.iot-a.eu/public/public-documents/d1.5/at_download/file 14
A reference model of IoT is
proposed by an FP7 project …
 What this model is helpful
for?
 Is it possible to provide
computerized support for
using this model?

1. Introductory examples – language definition
15
Language syntax definition (a toy language)
<postal-address> ::= <name-part> <street-address> <zip-part>
<name-part> ::= <personal-part> <last-name> <opt-suffix-part> <EOL>
|<personal-part> <name-part>
<personal-part> ::= <first-name> | <initial> "."
<street-address> ::= <house-num> <street-name> <opt-apt-num> <EOL>
<zip-part> ::= <town-name> "," <state-code> <ZIP-code> <EOL>
<opt-suffix-part> ::= "Sr." | "Jr." | <roman-numeral> | ""
<last-name> ::= [A-Z|a-z]
<first-name> ::= [A-Z|a-z]
personal-part
first-name
first-name-part initial
initial-part
last-name
suffix-part
house-num
street-num
Town-name
state-code
ZIP-code
component-p
postal-address
1,N
1,1
street-address-part
Zip-part
opt-apt-num

Agenda
16
4. Illustration
6. Conclusion
16

2. Abstraction levels – instanciation vs. generalisation
17
Object
CLASS
CLASS : a structure, a pattern
Object: identifier, values
• Attribute1
• Attribute2
• …
• "Value1"
• "Value2"
• …
CLASS B
• Attribute b1
• Attribute b2
• …
CLASS A
• Attribute a1
• Attribute a2
• …
InstanceOf
IsA

2. Abstraction levels – the meta-modeling language issue
18
A meta-model is described using a meta-modeling language
How to describe the meta-modeling language … a circular issue.

2. Abstraction levels – instantiation levels
• Meta-modeling framework by OMG : Meta Object Facility (MOF)
=> Strict instanciation
19

2. Abstraction levels – instantiation levels
20
• Instantiation levels in th Meta Object Facility (MOF) by OMG
Source: OMG

2. Abstraction levels – the “Powertype” approach
21
Cointe, Pierre: Metaclasses are first class: The ObjVlisp Model. SIGPLAN Notices.
22(12), 156–162 (1987).
 One of the earliest work on
these issues is by P. Cointe in
1987
Where does meta-modeling come from?
• Artificial intelligence (i.e. frame concept, Smalltalk, ObjVLisp)
• Databases (i.e. meta-database)
• Object oriented modeling (i.e. Classes are objects too => MetaClasses)
How to simultaneously create a generalization link AND an instantiation link
How to a define attributes to Classes
How to cross the strict instantiation levels

22
How to simultaneously create a generalization link AND an instantiation link
How to a define attributes to Classes
How to cross the strict instantiation levels
Pet
PetName
DateOfBirth
Owner
AnimalSpecies
SpecieName
AvgLifeSpan
Vaccinations
Diet
1 0..*
"Cat specie"
8
"Triple", "Leukaemia"
"Fish"
"Valentina"
8-sep-2008
"César"
Illustration (Source: Gonzalez-Perez & Henderson-Sellers, Metamodelling for software engineering, p.38)
How to model species-specific characteristics, e.g. Breed ?

23
Pet
PetName
DateOfBirth
Owner
CatSpecie
Breed
"Valentina"
8-sep-2008
"César"
Using specialization link to model
species-specific characteristics

24
Pet
PetName
DateOfBirth
Owner
AnimalSpecies
SpecieName
AvgLifeSpan
Vaccinations
Diet
1 0..*
"Cat specie"
8
"Fish"
"Valentina"
8-sep-2008
"César"
"Siamese"
CatSpecie
Breed
Specific representation
Class + Object

25
Pet
PetName
DateOfBirth
Owner
AnimalSpecies
SpecieName
AvgLifeSpan
Vaccinations
Diet
"Cat specie"
8
"Fish"
"Valentina"
8-sep-2008
"César"
"Siamese"
CatSpecie
Breed
PowerType and Clabject
Clabject
Powertype

26
Task
Start
End
Duration
Task Kind
TaskName
Purpose
MinCapLevel
"Code writing"
"To write code …"
1
"12-sep-07"
"18-sep-07"
7d
"C#"
CodeWritingTask
ProgLanguage
PowerType and Clabject for method engineering
Clabject
Powertype

2. Abstraction levels – deep instantiation
• Proposal:
– Add an indicator (potency) of the abstraction level at which the attribute
is to be instantiated
Task
+ TaskCategory1
+ TaskDuration0
CodeWritingTask
TaskCategory: Coding
T1
+ TaskDuration:10d
instance instance
27

Agenda
28
4. Illustration
6. Conclusion
28

29
(1)
(2)
(3)
Focus on 3 meta-modeling tools :

30
Other meta-modeling tools :
ADOxx Meta Modelling Platform
(http://www.adoxx.org/live/home)
(4)
(5)
(6) JastAdd (http://jastadd.org)
and
JastEMF
(https://code.google.com/a/eclipselabs.org/p/jastemf/)
for language engineering
(http://astreo.ii.uam.es/~jlara/metaDepth/)

31
Source http://download.eclipse.org/modeling/emf/emf/javadoc/2.9.0/org/eclipse/emf/ecore/package-summary.html
ECORE meta-modeling
framework (MOF based)

32
Source http://download.eclipse.org/modeling/emf/emf/javadoc/2.9.0/org/eclipse/emf/ecore/package-summary.html
ECORE meta-modeling
framework (MOF based)

33
MetaEdit (1)
• A meta-CASE tool for automatic customization of CASE tools
• Result of research project at Jyväskyla (K. Lyytinen, M. Rossi, S. Kelly et al. - 1990')
• Actually commercialized by MetaCASE
• Targeted towards Domain Specific Modeling
Meta-model definition Model definition
Generate
(Executable)
code

34
MetaEdit (2)
• Meta-modeling language: GOPRR
• Graph
• Object
• Property
• Relationship
• Role

35
MetaEdit (3)
• Meta-editing functionalities

36
MetaEdit (4)
• Customized editor

37
MetaEdit (5)
• MERL: A scripting language for generating code (i.e. flow of characters)

38
MetaEdit (6)
• Example in MERL for generating SQL code from E/R schema

39
Conceptbase (1)
• A deductive database supporting object-centered TELOS modeling language
• Result from research project DAIDA (J. Mylopoulos, M. Jarke et al. - 1990')
• Continued as an open source project (M. A. Jeusfled)
• Targeted towards Model Engineering
TELOS modeling language
• Any element (class, object, attribute, link, constraint … ) is internally represented as
a predicate in a deductive DB
• Attributes are first order objects i.e. they can have properties
• Attributes correspond to links between two elements
• Any object can be instantiated
 Multiple instantiation ('Paul' in 'Employee' and 'Paul' in 'Student')
 Multiple levels of instantiation
 Multiple Generalisation/Specialisation (i.e. heritage)
 Generalisation/Specialisation can co-exist with instantiation

40
Conceptbase (2)
• Model llustration
Person in Class with
attribute
ID: Integer;
name:String
end
Client isA Person with
attribute
reduction:Real
end
Employee isA Person with
attribute
function: String
end
Student isA Person with
attribute
age:Integer
end

41
Conceptbase (3)
• Multiple instantiation illustration
John in Student, Employee, Client with
ID
id_john: 44555
name
name_j: "John Legrand"
reduction
r1: 30.0
function
f1: "Sale manager"
age
age1: 24
end
Client with constraint
reduction_OK:
$ forall c/Client x/Real (c reduction x) ==> (x>=5.0) $
end
QueryClass AllPeople isA Person with
constraint
Allpeople_c: $ exists s/String (this nom s) $
end

42
AnimalSpecies in Class, MetaClass with
attribute
SpecieName: String;
AvgLifeSpan: Integer;
Vaccinations: String;
Diet: String
end
Pet in Class, MetaClass with
attribute
PetName: String;
PetDateOfBirth: String;
PetOwner: String
end
CatSpecie in AnimalSpecies, Class with
attribute
Breed: String
SpecieName
sn: "Cat specie"
AvgLifeSpan
als: 8
Vaccinations
v1: "Triple"; v2: "Leukaemia"
Diet
d1: "Fish“ end
MyCat in CatSpecie, Pet with
Breed
MyCatBreed:"Siamese"
PetName
p1: "Valentina"
PetDateOfBirth
d1: "08-Sep-2008"
PetOwner
pow: "César“ end
Conceptbase (3)
• Implementing Class attributes and the “PowerType” concept

43
Conceptbase (3)
• Implementing Class attributes and the “PowerType” concept

Agenda
44
4. Illustration
6. Conclusion
44

4. Meta-modeling illustration examples
45
The IoT ARM is exemplified with a
use-case scene : A load carrier is
equipped with sensors and can
communicate with other devices in
terms of wireless radio technology.
With this hardware, every load
carrier continuously measures its
environmental parameters and
sends all measurements via the
embedded event service to the
mobile phone of the truck driver.
Source : http://www.iot-a.eu/public/public-documents/d1.5/at_download/file
Exercise 1 : Internet of Things framework

46
“For example, Ted is a truck driver transporting highly sensitive orchids to a retail store. After loading the orchids
on his truck, he attaches an array of sensors to the load carriers in order to measure the temperature.
While having lunch, Ted forgot that by turning of the engine, air condition for the transported goods highly sensitive
orchids - shuts off, too. The temperature inside the truck starts rising and when it reaches a predefined critical
level inside, one of its sensors notices this and its node sends an emergency signal to Ted's IoT-Phone. On the
IoT-Phone's display, Ted can now see that the orchids are in danger so he rushes back to the vehicle and turns
the air condition on.
The IoT-Phone also keeps track of any alert messages it receives from the load carriers and saves this message
history for future inspection in a way that cannot be altered. When the truck reaches the retail store for delivery,
the sensor history is transferred to the store’s enterprise system and the sensors authenticate themselves as
being un-tampered.”
Exercise 1 : Internet of Things framework
Source : http://www.iot-a.eu/public/public-documents/d1.5/at_download/file
How to model this system using the IoT framework ?

47
Exercise 2: Method engineering
The design and development team in a company has defined a set of engineering methods
according to the nature of projects. Among others, and in the case of small size projects (number
of end-users <= 10, number of data tables <= 15, number of business process <=5), they have
defined the following simple method:
“The Requirements Engineering team (1 or 2 engineers) uses administrative documents and
conduct interviews with end-users to elicit general requirements, i.e a natural language
description of the required system (including NF requirements), and process definitions
represented as BMPN diagrams. The design team (2 to 4 engineers) uses the process models
together with the general description to sketch a system architecture, i.e. an UML class diagram,
use case diagrams and sequence diagrams. The development team (3 to 5 developers) build the
system according to the BPMN models and the system architecture, and deliver an operational
implementation composed of Java code for web services, a relational database and a workflow
script. The test team (2 to 3 persons) verify the quality of the system using a set of test cases
they have designed and in accordance with the NF requirements”.
How such method can be modeled using a
subset of SPEM and the MetaEdit+ workbench?

48
Exercise 3: Web questionnaires [Source : LWC2014 - Language Workbenches
Comparison, http://www.languageworkbenches.net/]
Forms-based software for data collection has found application in various areas, including scientific
surveys, online course-ware and guidance material to support the auditing process. As an overall
term for this kind of software applications we use the term "questionnaire". In this exercise, the goal
is to create a simple DSL, called QL, for describing questionnaires. Such questionnaires are
composed of sequential forms, each form is characterized by conditional entry fields and
(spreadsheet-like) dependency-directed computation. The following example presents a possible
textual representation of a simple questionnaire with only one form :
questionnaire HouseOwning
form Box1HouseOwning
{ hasSoldHouse: “Did you sell a house in 2010?” Boolean
hasBoughtHouse: “Did you by a house in 2010?” Boolean
hasMaintLoan: “Did you enter a loan for maintenance/reconstruction?” Boolean
if (hasSoldHouse)
{ sellingPrice: “Price the house was sold for:” money
privateDebt: “Private debts for the sold house:” money
valueResidue: “Value residue:” money(sellingPrice - privateDebt)
}
}

49
This simple form should generate into a GUI which allows the following user interaction:
Exercise 3: Web questionnaires [Source : LWC2014 - Language Workbenches
Comparison, http://www.languageworkbenches.net/]
How such language can be designed and implemented ?

Agenda
50
4. Illustration
6. Conclusion
50

6. Research challenges in meta-modeling
51
Behavior meta-modeling
• Behavior perspectives are generally missing in software engineering
meta-models
 Important knowledge about modeling languages is lacking
 Essential for tools designers and method engineers
 For a process modeling language, "behavior" perspective inquire
on its executable/operational semantics

6. Research challenges – behavior modeling
52
Research goal
• How to express the operational semantics for a modeling language?
• How to design and build enactment engines for a given modeling language?
 Maintainability and portability are central issues
[Source :T. Mayerhofer et al., “xMOF: Executable DSMLs Based on fUML”, SLE’2013 ]

53
• Context : MAP process modeling notation
Example

54
stopMapEnact()
selectCandidateSection()
stateMapInst.Old=Selected
stateMapInst.New=Running
M1:
Start
Map
Execution
notifyEndSection()
SectionInstance
Product Instance
SendCandidateSections()
EV4
EV5
EV9
MapActor
M2:Liste of candidates
M4:End of Section Execution
compute
Candidate
Sections()
IntentionInstance
Section
Application
Executor
updateIntention()
updateExecSection()
notifyEndExec()
invokeExec()
EV1
M3: Choice
C1: target intention= Stop
C1
MapInstance
startMapEnact()
stateMapInst.Old=Running
stateMapInst.New=Enacted
notifyEndMapEnact()
M6:Stop
stateSecInst.Old=Candidate
statSecInst.New=Selected
M8:Execution
status
computeCandidateSections()
stateSecInst.Old=Created
stateSecInst.New=Candidate
stateSecInst.Old=Selected
stateSecInst.New=Executed
EV3
EV6
EV8
EV12
Product
Manager
ImplemExe
SectionInstance
invokeExec()
notifyEndExec()
EV5
EV10 stateIntInst.Old=Created
stateIntInst.New=Realised
notifyProduct()
EV11
updateProduct()
M5: Product
M9: NewProduct
Situation
EV7
M7:Execution
ParametersstateImpl.Old=Created
stateImpl.New=Selected
C2
C2: target intention=/= Stop
newProductInstance()
EV2
M1bis:
Start MapInst
newMapInstance()
F1
F1: pour toute sectionInstance retournée par l'algorithme de calcul de candidates
Event-based behavior meta-modeling

55
• Other approaches
 KERMETA
For an example, see http://en.wikipedia.org/wiki/Kermeta

56
 xMOF (Mayerhofer et al., 2013)

Agenda
57
4. Illustration
6. Conclusion
57

6. Conclusion
58
• No well established and recognized standard for meta-modeling
• Tool support is complicated
• … OMG's MOF and IBM Eclipse emerge as "market" leaders for UML & Java
• MetaEdit: leader for DSM engineering tools
• Conceptbase: most powerful and theoretically sound modeling language
• Semantics for process meta-models still to be defined
• Beyond DSM, few empirical studies on meta-modeling applications

References
59
Links
• MOF homepage: http://www.omg.org/mof/
• ISO/IEC 24744:2007 (Software Engineering -- Metamodel for Development
Methodologies): http://www.iso.org/iso/catalogue_detail.htm?csnumber=38854
• OPEN (Object-oriented Process, Environment and Notation) :
http://www.open.org.au/
• Metacase: http://www.metacase.com/
• Conceptbase: http://conceptbase.sourceforge.net/
• Kermeta: http://www.kermeta.org
 S. Kelly et J.-P. Tolvanen, Domain-specific modeling:
enabling full code generation. Hoboken, N.J.: Wiley-
Interscience: IEEE Computer Society, 2008.
 C. Gonzalez-Perez et B. Henderson-Sellers,
Metamodelling for software engineering. Wiley
Publishing, 2008.
 Jeusfeld, M., Jarke, M., Mylopoulos, J.: Metamodeling
for method engineering. The MIT Press, (2009).

References
60
 Atkinson, C., Kühne, T., « The Essence of Multilevel Metamodeling », in: Gogolla, M. et
Kobryn, C. (éd.) «UML» 2001 — The Unified Modeling Language. Modeling Languages, Concepts,
and Tools. LNCS, vol.2185, pp. 19-33. Springer, 2001.
 C. Atkinson et T. Kühne, « Model-Driven Development: A Metamodeling Foundation », IEEE
Software, vol. 20, no 5, p. 36–41, 2003.
 C. Gonzalez-Perez et B. Henderson-Sellers, « A powertype-based metamodelling framework
», Software and Systems Modeling, vol. 5, no 1, p. 72-90, 2006.
 F. Jouault, F. Allilaire, J. Bézivin, et I. Kurtev, « ATL: A model transformation tool », Science of
Computer Programming, vol. 72, no 1-2, p. 31-39, 2008.
 A. Gargantini, E. Riccobene, et P. Scandurra, « A semantic framework for metamodel-based
languages », Automated Software Engineering, vol. 16, no 3-4, p. 415-454, 2009.
 M. Jarke, M. Jeusfeld, H. Nissen, C. Quix, et M. Staudt, « Metamodelling with Datalog and
Classes: ConceptBase at the Age of 21 », in Proceedings 2nd Int. Conf. on Object Databases
(ICOODB’09), M. Norrie et M. Grossniklaus, Éd. 2010, p. 95–112.
 J. de Lara et E. Guerra, « Deep Meta-modelling with MetaDepth », in Objects, Models,
Components, Patterns, J. Vitek, Éd. Springer Berlin Heidelberg, 2010, p. 1-20.
 B. Bryant, J. Gray, M. Mernik, P. Clarke, R. France, et G. Karsai, « Challenges and Directions
in Formalizing the Semantics of Modeling Languages », Computer Science and Information
Systems, vol. 8, no 2, p. 225-253, 2011.

References (cont’d)
61
 J. Sprinkle, B. Rumpe, H. Vangheluwe, et G. Karsai, « Metamodelling », in Model-Based
Engineering of Embedded Real-Time Systems, vol. 6100, H. Giese, G. Karsai, E. Lee, B. Rumpe,
et B. Schätz, Éd. Berlin/Heidelberg: Springer, 2011, p. 57-76.
 A. El Kouhen, C. Dumoulin, S. Gerard, et P. Boulet, « Evaluation of Modeling Tools
Adaptation », 2012. Available at http://hal.archives-ouvertes.fr/hal-00706701
 B. Henderson-Sellers, O. Eriksson, C. Gonzalez-Perez, et P. J. Ågerfalk, « Ptolemaic
Metamodelling? The Need for a Paradigm Shift », in Progressions and Innovations in Model-
Driven Software Engineering:, V. G. Díaz, J. M. C. Lovelle, B. C. P. García-Bustelo, et O. S.
Martínez, Éd. IGI Global, 2013, p. 90-146.
 O. Eriksson, B. Henderson-Sellers, et P. J. Ågerfalk, « Ontological and linguistic
metamodelling revisited: A language use approach », Information and Software Technology, vol.
55, no 12, p. 2099-2124, 2013.
 C. Bürger, S. Karol, C. Wende, et U. Aßmann, « Reference Attribute Grammars for
Metamodel Semantics », in Software Language Engineering, vol. 6563, B. Malloy, S. Staab, et M.
van den Brand, Éd. Berlin/Heidelberg: Springer, 2011, p. 22-41.
 T. Mayerhofer, P. Langer, M. Wimmer, et G. Kappel, « xMOF: Executable DSMLs Based on
fUML », in Software Language Engineering (SLE’13), M. Erwig, R. F. Paige, et E. V. Wyk, Éd.
Springer International Publishing, 2013, p. 56-75.
 B. Neumayr, M. A. Jeusfeld, M. Schrefl, et C. Schütz, « Dual Deep Instantiation and Its
ConceptBase Implementation », in Advanced Information Systems Engineering, M. Jarke, J.
Mylopoulos, C. Quix, C. Rolland, Y. Manolopoulos, H. Mouratidis, et J. Horkoff, Éd. Springer
International Publishing, 2014, p. 503-517.
 R. F. Paige, D. S. Kolovos, et F. A. C. Polack, « A tutorial on metamodelling for grammar
researchers », Science of Computer Programming, vol. 96, Part 4, p. 396-416, 2014.

THANK YOU FOR YOUR ATTENTION !
62
Questions? Comments? Insights?
Homepage
http://fr.linkedin.com/pub/sa%C3%AFd-assar/4/68a/66a
http://fr.slideshare.net/SaidAssar/
https://twitter.com/Said_Assar/
http://www-public.it-sudparis.eu/~assar/

Meta-modeling: concepts, tools and applications

More Related Content

What's hot

Viewers also liked

Similar to Meta-modeling: concepts, tools and applications

More from Saïd Assar

Recently uploaded

Meta-modeling: concepts, tools and applications