models 2010 - a systematic review on the definition of uml profiles
TRANSCRIPT
A Systematic Review onthe Definition of UML Profiles
Jesús PardilloUniversity of Alicante, Spain
✗
✔
Jesús Pardillo: A Systematic Review of UML Profiles. MODELS, Oslo, 6th Oct. 2010 2
UML Profiles
Jesús Pardillo: A Systematic Review of UML Profiles. MODELS, Oslo, 6th Oct. 2010 3
Some UML Profiles
• Softw. Requirements, Quality, Rationale
• Data Mining: Time Series, Clustering, etc.
• Softw. Personalisation
• Dataflows
• Databases
• Security
Jesús Pardillo: A Systematic Review of UML Profiles. MODELS, Oslo, 6th Oct. 2010 4
Some Pitfalls
• Lack of Formal Semantics
• Lack of Expressiveness
• Notational
• Syntactical
• Large UML metamodel
• Lack of Tooling (OCL, XMI, Querying)
• Lack of Customisability
• Trilingualism
• Lack of Completeness and Consistency
• Non-uniform Access
������� ������������������ � ��
��
��
��
���� ���� ���� ���� ���� ���� ���� ���� ���� ���� ���� UML 2.0
Publication Trend
‣ Raw Numbers:
‣ 39 Articles
‣ 406 Stereotypes
‣ 46 Distinct Metaclasses
‣ 300 Tags (~0.7 Tags / Stereotype)
‣ Median By Profile:
‣ 9 Stereotypes
‣ 4 Distinct Metaclasses
‣ Domains:
arch, bp, db, reqs, sec, tmp, ui, web, ...
5Jesús Pardillo: A Systematic Review of UML Profiles. MODELS, Oslo, 6th Oct. 2010
!"#$%&'()*+
!'',-$./,*$!'()*+
!'()*+
!'(/0$1+
!'(/0$12&%(()*+
!'$)%+
!%(3&'$+
!##)'0&()*+
!##)'0&()*45&##+
6,7&/0)%+
45&##+
45&##08,%+4)55&")%&()*+
4)99,*$+4)9-)*,*$+
4)*#$%&0*$+
4)*$%)5:5);+
<&$&=1-,+<,-,*>,*'1+
<,-5)19,*$+
.5,9,*$+
.*?9,%&()*+
.*?9,%&()*@0$,%&5+A*3)%9&()*:5);+
A*#$&*',B-,'08'&()*+
A*$,%&'()*+
A*$,%3&',+
4)99?*0'&()*2&$7+
C,##&D,+
C)>,5+E&9,>.5,9,*$+
E)>,+F"G,'$:5);+
F"G,'$E)>,+
F-,%&()*+
2&'H&D,+
2)%$+
2%090(/,=1-,+
2%)-,%$1+
I,5&()*#70-+
B$&$,C&'70*,+
B$%?'$?%,>!'(/0$1E)>,+
B$%?'$?%,>45&##08,%+
=09,./,*$+
=%&*#0()*+
J#&D,+
J#,4&#,+
6Jesús Pardillo: A Systematic Review of UML Profiles. MODELS, Oslo, 6th Oct. 2010
Pro
file
s by
Met
acla
ss
!"#$%#&'()*+)
,-.-&/."0%1&'()2+)
3'&4.'&'()2+)
!"#51'()2+)
61-&7."#51'()8+)
49.''&'()*:+)
;&<95=>&1-'()?+)
45><51&1-'()?+)
45><5'%-&,-7@"-@7&'()
2+)
!@A%9%.7=451'-7@"-'()
2+)
7Jesús Pardillo: A Systematic Review of UML Profiles. MODELS, Oslo, 6th Oct. 2010
... and by Language Unit
!"
#!"
$!"
%!"
&!"
'!"
(!"
)*+,-."
/01..2."
)*+34+2."
5-6271*+,-."
)8940417:/,-.678*6."
;6162<1*=4-2."
/,>?,.462;678*6872."
@2?0,:>2-6."
A.2/1.2."
/,>?,-2-6."
B962-C2C" A<D"
8Jesús Pardillo: A Systematic Review of UML Profiles. MODELS, Oslo, 6th Oct. 2010
Extended Metaclasses by Language Unit
‣ Iconography by Stereotype:
‣ Class: 40%
‣ Properties: 28%
‣ Associations: 17%
‣ Proportions By Profile:
‣ ½ with Iconography(¼ Icon by Stereotype)
‣ ¾ with Tagging(⅓ Tag by Stereotype)
‣ ½ with (visible) Constraints(?? Constraint by Stereotype)
9Jesús Pardillo: A Systematic Review of UML Profiles. MODELS, Oslo, 6th Oct. 2010
Iconography, Tagging, and Constraining
Presentation Trend
10Jesús Pardillo: A Systematic Review of UML Profiles. MODELS, Oslo, 6th Oct. 2010
!"
#"
$"
%"
&'"
&("
&)"
'&"
'*"
'+"
#!"
##"
,!!-,!&" ,!'-,!#" ,!*-,!(" ,!$-,!+" ,!)-,!%"
./01234/5"6.7"
89:5"687"
;/430<=/25"6;7"
><?032"@/03ABCD5"6>7"
EF3G?305"6E7"
H22"9<AD4?3FA4D5"697"
I?<J2/D"
‣ Behavioural Profiling:
✓Profile Diagram
✓Formal Semantics
‣ Structural Profiling:
✓All Constraints
✓Metamodel
✓Template
11Jesús Pardillo: A Systematic Review of UML Profiles. MODELS, Oslo, 6th Oct. 2010
Presentation Patterns by Concern
12Jesús Pardillo: A Systematic Review of UML Profiles. MODELS, Oslo, 6th Oct. 2010
Summary
‣ Less UML profiles are published.
‣ Metaclass extension is heterogeneousand biased: Class, Property, and Association.
‣ Occasional iconography mainly on:Class, Property, and Association.
‣ Tagging is more popular than iconography.
‣ Constraints are neglected.
‣ Defined informallybut slightly regular: structures vs. behaviour.
13Jesús Pardillo: A Systematic Review of UML Profiles. MODELS, Oslo, 6th Oct. 2010
Research Questions
‣ Are UML profiles well understood?
‣ Are formal methods useful for profiling?
‣ Are semantics of UML being neglected?
‣ Is “MOF” enough?
‣ Is UML being profiled for notational reasons?
‣ Is iconography enough?
Jesús Pardillo. A Systematic Review of UML Profiles. MODELS, Oslo, 6th Oct. 2010 14
My Educated Guesses
‣ Profiling for:
✴ Specialisation
✓ Visualisation
✓ Annotation
Jesús Pardillo. A Systematic Review of UML Profiles. MODELS, Oslo, 6th Oct. 2010 15
Profiling for Specialisation
Jesús Pardillo. A Systematic Review of UML Profiles. MODELS, Oslo, 6th Oct. 2010 16
Profiling for Visualisation
Jesús Pardillo. A Systematic Review of UML Profiles. MODELS, Oslo, 6th Oct. 2010 17
UMLUML Profiles
Profiling for Annotation
A Systematic Review onthe Definition of UML Profiles
Thank you very much for your attention
• Further reading:Domain-specific language modelling with UML profiles by
decoupling abstract and concrete syntaxes (JSS’10)
‣ Are UML profiles well understood?
‣ Are formal methods useful for profiling?
‣ Are semantics of UML being neglected?
‣ Is “MOF” enough?
‣ Is UML being profiled for notational reasons?
‣ Is iconography enough?
Jesús PardilloUniversity of Alicante, Spain