Modelling and Analysis of CSCW systems: An Ontology-driven Engineering Approach Supervisors: Dr....
-
Upload
allen-chivers -
Category
Documents
-
view
214 -
download
1
Transcript of Modelling and Analysis of CSCW systems: An Ontology-driven Engineering Approach Supervisors: Dr....
Modelling and Analysis of CSCW systems:An Ontology-driven Engineering Approach
Supervisors: Dr. José Luis Garrido Bullejos
Thesis defense
Departamento de Lenguajes y Sistemas InformáticosUniversidad de Granada
Candidate: Manuel Noguera García
Dra. María V. Hurtado Torres
2
Outline
• Introduction Motivation Goals Foundations Intended Approach
• AMENITIES• Ontology-driven modelling and analysis of CSCW systems
Ontology implementation techniques Modular design scheme to ontology development Ontology-based reasoning
• Applications of the proposal• Conclusions & Future Work
3
Motivation• Increment in the complexity of the tasks
to be carried out with computing systems Involvement of several people/organizations in their
accomplishment Incorporate collaboration capabilities in the system used
• Computer-supported Cooperative Work (CSCW) systems:
Intended to help people work efficiently Strongly influenced by social (human) aspects Require (as much as possible) complete, clearly-defined,
easy-to-manage system models that• cover both structure and behavior• offer general/abstract views of the system to discuss with
collaborating stakeholders
A great deal of effort in specification
CSCW
4
Motivation
• Models usually focus on selected particular aspects Several models are needed for the whole system Scattering of information & design decisions Unnoticed inconsistencies between models
• Semantics is often unclear or too informal Misunderstandings Reduce potential of knowledge shared Difficult communication, coordination, and thus, collaboration
between partners
CSCW
5
Goals of the thesis• General goal
Specify collaborative systems through models that:• Capture both structure and behavior• Can be obtained in a systematic manner• Have a clearly-defined semantics• Allow consistency checks to be carried out• Provide a cohesive representation of the system
• Secondary goals Provide a set of techniques to systematically represent common
conceptual modelling constructs Apply the proposed methods to different domains Develop a tool that assists analysts in the description of CSCW
system models
6
FoundationsModel-driven Engineering (MDE) approaches as new
paradigms to System and Software Engineering:
• Extensive use of models to system development
• Aim: Raise the abstraction level of models Foster discussion with stakeholders Separate business logic from implementation issues Enable the implementation of a business logic across different
technological platforms Computation and Technology Independent Models
• Adopt UML as the reference modelling notation
• Benefit: User-friendly, intuitive models
• Drawback:• Lack formal and complete model theoretic semantics to carry out
automated reasoning and validation
• Spread of information and design decisions across different models (UML 13 different diagrams for system architecture)
MDE
7
FoundationsOntology:• Originally a branch of Metaphysics (or Philosophy)
• Specialized meaning in Computer Science: Formal specifications about a domain It is possible
to talk of an ontology or several ontologies An ontology = classes (a.k.a. concepts) +
relationships (a.k.a. properties and slots) + restrictions on these relationships (a.k.a. facets)
Benefit:• Enable logic-based automated reasoning and
consistency checks on the models Drawback:
• Lack user-friendly notation not suitable to discussion with stakeholders
• Focus on the structure of concepts rather than the processes to describe a domain absence native support to describe behaviour
Ontologies
8
Intended Approach
• Ontology Driven Engineering (ODE) Combined approach of MDE and formal ontologies Models are formally captured in underlying ontologies Take advantage of the benefits of both technologies:
• High-abstraction level and user-friendly models to discuss with stakeholders• Formal specifications about a domain to carry out consistency checks and
infer implicit knowledge
• Approach: Devise and apply an ODE process to the modelling and analysis of CSCW systems may help improve their specification process
MDE Ontologies+ =MDE Ontologies ODE CSCWfor
9
Outline
• Introduction Motivation Goals Foundations Intended Approach
• AMENITIES• Ontology-driven modelling and analysis of CSCW
systems Ontology implementation techniques Modular design scheme to ontology development Ontology-based reasoning
• Applications of the proposal• Conclusions & Future Work
10
Starting point: AMENITIES [Garrido 2005]
• “A MEthodology for
aNalysing and desIgning
collaboraTIve systEmS” Core of the methodology:
Cooperative Model (COMO)
Requirement Models
UML Use CaseApplied
Ethnography
Cooperative Model(COMO-UML)
Software DevelopmentModels (UML)
Formal Model
UML Statecharts
UMLDiagrams
Refine
(Coloured Petri Nets)
AdditionalRequirements
Revise
Revise
Analyse Develop
Model Requirements
FunctionalRequirements
Organizational View(Organization, Roles,…)Organizational View
(Organization, Roles,…)
Interaction View(Protocols, Artefacts,…)Interaction View
(Protocols, Artefacts,…)
Cognitive View(Tasks, Actions,…)Cognitive View
(Tasks, Actions,…)
Information View(Documents, Messages,…)Information View
(Documents, Messages,…)
Cooperative Model(COMO-UML)
11
Starting point: AMENITIES [Garrido 2005]
• Conceptual framework Domain vocabulary Main entities in a
collaborative system Described using natural
language and UML
12
Starting point: AMENITIES [Garrido 2005]
• Views of the system
Organization diagrams Role diagrams Task diagrams
• Make use of and extend UML lack a formal semantics to carry out automated consistency checks or reasoning
• Approach: Representation in an ontology language
13
Outline
• Introduction Motivation Goals Foundations Intended Approach
• AMENITIES• Ontology-driven modelling and analysis of CSCW
systems Ontology implementation techniques Modular design scheme to ontology development Ontology-based reasoning
• Applications of the proposal• Conclusions & Future Work
14
Ontology Implementation
• First step: Language election
• Candidates languages: KIF, LOOM, RDF, OWL...
• Choice: OWL-DL (Web Ontology Language – Description Logics) W3C standard Machine-processable descriptions that foster interoperability between
software agents Plenty of related technologies Reasoning capabilities based on Description Logics
15
Ontology Implementation• Next step: Representation of the AMENITIES conceptual framework in
OWL
• Process: define classes in the ontology arrange the classes in a taxonomic (subclass–superclass) hierarchy define relationships describe allowed values for these relationships
• Guidance: UML class diagram representing the conceptual framework of the methodology
• Method (usual in the bibliography): Classes Concepts Associations Properties Aggregations part_of / has_part properties Is_a Subconcepts Multiplicity Cardinality restrictions
16
Ontology Implementation
17
Ontology Implementation. Limitations of the OWL language
• Limitations: • Adopted solutions (design patterns):
Cardinality restrictions on transitive properties
No native support for processes (focus put on structure rather than behaviour)
Transitive superproperties
Extra classes and relationships
Inability to representn-ary relationships Reified relationships
18
• In OWL, all relationships are binary Impossibility to represent n-ary relationships
• Usual and useful conceptual modelling construct• In AMENITIES: e.g., transitions between roles
Situation: An actor playing a role may start playing another one under certain conditions
Three participants in the relationship:1. Source role2. Destination role3. Guard to be satisfied
• Design pattern: A new class whose instances represent instances of the relationship “N” new functional relationships, i.e., as much as classes participating
in the n-ary relationship
Ontology Implementation. Representation of n-ary relationships
Organization Branch3..4
[Tel
ler?
]
Role Teller2
Role BankManager1
Role HeadOfRisk1
[HeadOfRisk?]
[BankManager?]
[Abs
ent(
Ban
kMan
ger)
]
19
Ontology Implementation. Representation of n-ary relationships
• Subclasses of a new class “Reified_Relation” Semantic information for ontology editors, software agents and system analysts
reified relation
functional properties
source role
destination role
guard to evaluate
20
Ontology Implementation. Limitations of the OWL language
• Limitations: • Adopted solutions (design patterns):
Cardinality restrictions on transitive properties
No native support for processes (focus put on structure rather than behaviour)
Transitive superproperties
Extra classes and relationships
Inability to representn-ary relationships Reified relationships
21
Ontology Implementation. Cardinality restrictions on transitive properties
• Certain relations exhibit an intrinsic transitive nature (e.g. aggregations):
if A has_part B, B has_part C A has_part C (could be automatically inferred)
In UML it is not possible to specify transitivity (in OWL it is) Useful to relate concepts
• E.g. CSCW_Systems are composed of Organizations, Organizations are composed of Roles, etc. which Roles make up a particular CSCW_System?
• Additionally, convenience of defining certain cardinality restrictions Organizations should be composed of at least one Role Groups should be composed of at least two Actors
• Cardinality + transitive is forbidden in OWL for decidability issues
22
Ontology Implementation. Cardinality restrictions on transitive properties
• Design pattern: A new relationship,
“superproperty” with the same intended meaning is defined, e.g., comprises for has_part
Transitivity is declared on the superproperty (i.e., comprises)
Cardinality restrictions are placed on the subproperty (i.e., on has_part, for example)
23
Ontology Implementation. Limitations of the OWL language
• Limitations: • Adopted solutions (design patterns):
Cardinality restrictions on transitive properties
No native support for processes (focus put on structure rather than behaviour)
Transitive superproperties
Extra classes and relationships
Inability to representn-ary relationships Reified relationships
24
Ontology Implementation. No native support for processes
• Ontology languages lack of native support to represent processes
• Essential in CSCW system specification to produce helpful models: Tasks and activities are to be arranged in ordered sequences. Flows of
activities may fork, join, jump backwards/forwards, etc. Tasks and activities are to be reused in the same or different workflows
A task/activity should be considered irrespective of its actual execution
• Unsupported in the AMENITIES conceptual framework UML activity diagrams are subsequently used in the methodology, but
order between activities is not explicitly addressed
25
Ontology Implementation. No native support for processes
• Solution set of extra classes and relationships: The execution of each task is modelled as a sequence of steps
(new classes) Each step (but the final_step) may be followed_by (new
relationship) one or more steps At each step it may take place an activity, an action, a workflow
fork, join, etc.
valuationReport [Finished]
appraiser:value
headOfRisk:collectApplicantData
first_step_1
fork_step_1
followed_by
action_step_1
followed_byfollowed_by
inf_object_step_1
activity_step_1
join_step_1
followed_by
followed_by
followed_by
performs
performs
is_produced
OWL Description UML Activity Diagram
26
Ontology Implementation. No native support for processes
27
Ontology Implementation. No native support for processes
• How about guards? Rule over the transition between two activities (steps actually...) 3-ary relationship: source step, destination step and the guard to be evaluated Reified relation design pattern: followed_by relationship reified in a class
debtReportStatus
headOfRisk:prepareDocuments
[Refusal][Hesitant]
[Passed]bankManager+headOfRisk:decideConcession
decision_step_1
followed_by_relation_1
guard_hesitantactivity_step_1
followed_by
following_step evaluates
followed_by_relation_2
activity_step_2
following_step
guard_passed
evaluatesfollowing_step
final_step_1followed_by_relation_2
guard_refusalfollowed_by
followed_byfollowing_step
evaluates
• Complicated?
• How about the semantics? Structurally the same as the UML metamodel for Activity Diagrams One-to-one match
• Activity/Action Class Activity/Action Concept
• ActivityNode/ActionNode ActivityStep/ActionStep
• ActivityEdge (reified) Followed_by_Relation
• ControlNode (InitialNode, FinalNode, ForkNode,...) Control_Flow_Step (First_Step, Final_Step, Fork_Step,...)
28
29
AMENITIES conceptual framework
classes added forprocess modeling
30
Outline
• Introduction Motivation Goals Foundations Intended Approach
• AMENITIES• Ontology-driven modelling and analysis of CSCW
systems Ontology implementation techniques Modular design scheme to ontology development Ontology-based reasoning
• Applications of the proposal• Conclusions & Future Work
31
Modularity in Ontology Design
A modular, multi-tier scheme for ontology design to simplify:
• Ontology refinement/updates Modifying a module should not lead to modifications in parts of the
ontology that are not conceptually related• Integration with ontologies from other organizations
Relationships between different ontology modules are controlled no unexpected consequences
• Partial reuse Reuse only the relevant part/module of an ontology
32
Amenitiesconceptual framework
ontologydocument
Modularity in Ontology Design.Multi-tier Scheme
Amenities-basedapplication ontology
document forenterprises
Amenities-basedapplication ontology
document forC&C Valuation Office
Amenities-basedapplication ontology forJohn F. Kennedy aiport
Amenities-basedapplication ontology
document forairports
Amenities-basedapplication ontology
document forOxford University
Amenities-basedapplication ontology
document forOxford University
Amenities-basedapplication ontology
document forNotary’s Offices
Amenities-basedapplication ontology
document forKlimt Notary’s Office
Amenities-basedapplication ontology
document foruniversities
Ground levelapplication ontologies
First levelapplication ontologies
Amenities-basedapplication ontology
document forBank of Santander
Amenities-basedapplication ontology
document forBranch Office nº 15
At the top levelAMENITIES conceptual framework
domain vocabulary of CSCW systems
At the next levelsome instances or more refined classes
for more specific domains would be described
Finally, more specific ontologiesrelated to particular
collaborative environments Amenities-basedapplication ontology
document forBranch Office nº 27
Domain ontology
33
Outline
• Introduction Motivation Goals Foundations Intended Approach
• AMENITIES• Ontology-driven modelling and analysis of CSCW
systems Ontology implementation techniques Modular design scheme to ontology development Ontology-based reasoning
• Applications of the proposal• Conclusions & Future Work
34
Analysis of the Specifications. Ontology-based Reasoning
• Automated reasoning procedures allow Help design and maintain sound
ontologies by:• Detecting unnoticed logic
consequences or inconsistencies
• Inferring non-explicit knowledge
• Ontologies drive the specification and analysis of the CSCW system
Coop-Task mortgageGranting
[Refusal]
valuationReport[Finished]
appraiser:value
headOfRisk:feasibilityStudy
payroll
unpaids
accounts
debtReport
debtReportStatus
bankManager+headOfRisk:decideConcession
headOfRisk:prepareDocuments
[Refusal]
[Hesitant]
[Passed]
[Passed]
draft
bankManager:giveApproval
draft[Signed]
titleDeed[Unsigned]
teller:openAccount
headOfRisk:createDeed
notary+bankManager+client:agreeAndSign
titleDeed[Signed]
Protocol conversational-communication Requirements { face-to-face
shared-workspace}
Protocol negotiation-communication Requirements{ face-to-face
shared-workspace}
headOfRisk:collectApplicantData
•In the mortgageGranting task:
•decideConcession is followed_by prepareDocuments
•prepareDocuments is followed_by giveApproval
then it can be inferred
•givesApproval succeeds decideConcession
•In the mortgageGranting task:
•Individual(decideConcession) type (Risk)
•Individual(giveApproval) type (Supervision)
RISK
SUPERVISION
“A subactivity or action classified as risky must always be followed by a supervision subactivity”
Rule satisfied
Two additional relationships, precedes and succeeds, can be defined as the transitive closure of followed_by and comes_after, respectively
•Reasoning on Transitivity
35
Outline
• Introduction Motivation Goals Foundation Intended Approach
• AMENITIES• Ontology-driven modelling and analysis of CSCW
systems Implementation Modular design Ontology-based reasoning
• Applications of the proposal• Conclusions & Future Work
36
Applications. Interaction observation systems: the case of COLLECE [Bravo 2007]
37
Applications. Interaction observation systems: the case of COLLECE [Bravo 2007]
38
Applications. Design of Case-Based Reasoners (CBR)
39
Visual Tool for Ontology Edition
• OWL syntax is rather verbose ontology edition a cumbersome task
• Diagrammatic representations help provide a general view of ontologies at a glance
• Aim: Facilitate ontology edition in a modular manner
40
Visual Tool for Ontology Edition
41
Visual Tool for Ontology Edition
42
Conclusions• We have extended and formalized the AMENITIES conceptual
framework in a formal ontology
Several techniques and design patterns have been provided for systematically representing usual conceptual modelling constructs in OWL
We have provided a set of classes and relationships that enable the description of workflows in the OWL language
We have defined a mapping between the entities of the UML metamodel for activity diagrams and a set OWL constructs to describe workflows without information lost
• We have devised a modular approach for the construction of collaborative system ontologies.
The resulting ontologies are formal underlying CIM’s for an ontology-driven engineering approach to the development of collaborative systems
43
Conclusions• We have presented a formalization of collaborative-system models
by means of OWL ontologies, that facilitates: Early detection of inconsistencies and/or meaningless concept
structures Inference of non-explicitly declared facts Further reasoning capabilities on the order of the activities in a workflow
• We have devised an interaction observation system that makes use of ontologies to obtain analysis descriptors
• We have made used of ontologies to model the structure of case descriptions to be subsequently used by CBR systems in solution searching and retrieval
• Finally, we have started the development of a visual ontology editor intended to guide the designer in the modular construction of ontologies
44
Future Work
• Definition of a service ontology Service Oriented Computing Transition from computation-independent to platform-independent
models
• Inclusion of goals in the ontology Most of groupware fails in goal analysis In CSCW different entities have different goals Goals affect and even conflict one another
45
Selected publications1. Noguera, M., Hurtado, M.V. et al.: “Ontology-driven Analysis of UML-Based Collaborative Processes using OWL-DL and CPN”.
Science of Computer Programming, (in press), 2009
2. Duque, R., Noguera, M. et al.: “Construction of interaction observation systems for collaboration analysis in groupware applications”. Advances in Engineering Software, Elsevier, 2009. doi:10.1016/j.advengsoft.2009.01.028
3. Penichet, V.M.R., Rodríguez, M.L., Lozano, M.D., Garrido, J.L., Gallud, J.A., Noguera, M., Tesoriero, R., Hurtado, M.V.: “Extending and Supporting Featured User Interface Models for the Development of Groupware Applications”. Journal of Universal Computer Science, Vol.14, No. 19, 3053-3070, 2008
4. Garrido, J.L., Noguera, M. et al.: “Definition and Use of Computation Independent Models in an MDA-Based Groupware Development Process”. Science of Computer Programming, Vol. 66, nº1, 25-43, 2007
5. Duque, R., Rodríguez, M.L., Hurtado, M.V., Noguera, M., Bravo, C.: “An Architecture to Integrate Automatic Observation Mechanisms for Collaboration Analysis in Groupware”. VII International Workshop on System/Software Architectures, OTM Workshops, Monterrey, México. Springer-Verlag, LNCS 5333, 354 – 363, 2008
6. Rodríguez, M.L., Garrido, J.L., Hurtado, M.V., Noguera, M., Hornos, M.J.: Design Guidelines for the Construction of User Interfaces for Collaborative Applications: A Model-Based Approach. Springer, 2009
7. Garrido, J.L., Hurtado, M.V., Noguera, M., Zurita, J.M.: “Using a CBR Approach based on Ontologies for Recommendation and Reuse of Knowledge Sharing in Decision Making”. 8th International Conference on Hybrid Intelligent Systems (HIS 2008). IEEE Press, 2008, 837-842
8. Duque, R., Noguera, M., Bravo, C., Garrido, J.L., Rodríguez, M.L.: “Construcción de un Sistema de Observación de la Interacción para Entornos CSCW”. IX Congreso de Interacción Persona Ordenador (AIPO) [Interacción’2008], Albacete, España, Thomsom Scientific. (2008 Jesús Lorés Award)
9. Noguera, M., Hurtado, M.V., Rodríguez, M.L., Chung, L., Garrido, J.L.: “Description of Collaborative Processes using OWL-DL”. The 2007 International Conference on Software Engineering Research and Practice, Las Vegas, Estados Unidos. CSREA Press, 574-580, 2007
10. Rodríguez, M.L., Garrido, J.L., Hurtado, M.V., Noguera, M.: “An Approach to the Model-based Design of Groupware Multi-user Interfaces”. 13th International Workshop on Groupware (CRIWG 2007), Bariloche, Argentina. Springer-Verlag, LNCS 4715, 157-164, 2007
11. Hurtado, M.V., Noguera, M., Rodríguez, M.L., Garrido, J.L., Chung, L.: “An Ontology-based Approach to the Modeling of Collaborative Enterprise Processes: Dynamic Managing of Functional Requirements”. Second International Conference on Evaluation of Novel Approaches to Software Engineering, Barcelona, España. INSTICC Press. 87-94, 2007
12. Noguera, M., Hurtado, M. V., Garrido, J.L.: “An Ontology-Based Scheme Enabling the Modeling of Cooperation in Business Processes”. International Workshop on Modeling Inter-Organizational Systems, OTM Workshops, Montpellier, Francia. Springer-Verlag, LNCS 4277, ISSN: 0302-9743, 863 – 872, 2006
17 additional publications in refereed journals and conferences...
Modelling and Analysis of CSCW systems:An Ontology-driven Engineering Approach
Supervisors: Dr. José Luis Garrido Bullejos
Thesis defense
Departamento de Lenguajes y Sistemas InformáticosUniversidad de Granada
Candidate: Manuel Noguera García
Dra. María V. Hurtado Torres
Modelado y Análisis de Sistemas CSCW siguiendo un enfoque de
Ingeniería Dirigida por Ontologías
Directores: Dr. José Luis Garrido Bullejos
Tesis Doctoral
Departamento de Lenguajes y Sistemas InformáticosUniversidad de Granada
Doctorando: Manuel Noguera García
Dra. María V. Hurtado Torres
48
AMENITIES 2. New conceptual framework
49
UML Metamodel for Activity Diagrams
50
Reasoning on Activity Ordering
ObjectPropertyAssertion(amenities:followed_by step_w fwd_by_x)
ObjectPropertyAssertion(amenities:following_step fwd_by_x step_x)
ObjectPropertyAssertion(amenities:precede step_w step_x)
ObjectPropertyAssertion(amenities:followed_by step_x fwd_by_y)
ObjectPropertyAssertion(amenities:following_step fwd_by_y step_y)
ObjectPropertyAssertion(amenities:precede step_x step_y)
ObjectPropertyAssertion(amenities:followed_by step_y fwd_by_z)
ObjectPropertyAssertion(amenities:following_step fwd_by_z step_z)
ObjectPropertyAssertion(amenities:followed_by step_w fwd_by_x)
ObjectPropertyAssertion(amenities:performs
step_w decideConcession)
ObjectPropertyAssertion(amenities:performs
step_x prepareDocuments)
ObjectPropertyAssertion(amenities:generate
step_y draft)
ObjectPropertyAssertion(amenities:performs
step_z giveApproval)
ClassAssertion(decideConcession
amenities:Risky)
ClassAssertion(giveApproval
amenities:Supervision)
declaration of the typesof the activities
specification of the control flow
between activities
particular actions/activities to be performed in every step
inferred
ObjectPropertyAssertion(amenities:precede step_w step_x)
ObjectPropertyAssertion(amenities:precede step_x step_y)
ObjectPropertyAssertion(amenities:precede step_y step_z)
ObjectPropertyAssertion(amenities:precede step_w step_z)
inferred
inferred
inferred
summary of the full reasoning process
52
Metamodelling
53
Starting point: AMENITIES [Garrido 2005]
• “A MEthodology for
aNalysing and desIgning
collaboraTIve systEmS” Core of the methodology:
Cooperative Model (COMO)
Makes use of and extends UML
Lacks of formal semantics to
carry out consistency checks or
reasoning
Requirement Models
UML Use CaseApplied
Ethnography
Cooperative Model(COMO-UML)
Software DevelopmentModels (UML)
Formal Model
UML Statecharts
UMLDiagrams
Refine
(Coloured Petri Nets)
AdditionalRequirements
Revise
Revise
Analyse Develop
Model Requirements
FunctionalRequirements
Organizational View(Organization, Roles,…)Organizational View
(Organization, Roles,…)
Interaction View(Protocols, Artefacts,…)Interaction View
(Protocols, Artefacts,…)
Cognitive View(Tasks, Actions,…)Cognitive View
(Tasks, Actions,…)
Information View(Documents, Messages,…)Information View
(Documents, Messages,…)
Cooperative Model(COMO-UML)
55
Goals of the thesis• General goal:
Use of ontologies so as to obtain a formal underlying representation of the
AMENITIES collaborative model, and thus, set the basis for the adoption of ODE
approaches in the construction of CSCW systems
• Intermediate goals Analyse the state of the art in conceptual modelling and ontology specification
languages to represent domain knowledge Define an ontology for the conceptual framework proposed in the AMENITIES
methodology enabling to:• Carry out consistency checks• Capture both structure and behavior of a collaborative system
Provide a set of ontology design patterns intended to represent common conceptual modelling constructs and/or avoid some limitations in its use
Illustrate the benefits of the use of ontologies proposed by means of automated reasoning to detect possible inconsistencies or infer knowledge not explicitly declared
Apply the proposed techniques on real case studies Develop a tool to enable the visual edition of ontologies that assists analysts in the
adoption of an ODE approach to system construction