2CL Protocols: Interaction Patterns Specification inCommitment Protocols
Elisa Marengo
Universita degli Studi di TorinoScuola di Dottorato in Scienze ed Alta Tecnologia
Dottorato in Informatica
Ph.D. Thesis Defence - Torino, October 19, 2012
Advisor: Prof. Matteo Baldoni - Prof. Cristina BaroglioElisa Marengo (UNITO) 2CL Protocols October 19, 2012 1 / 41
Interaction Protocols
Representation of Complex Systems
Current trends:
Advent of distributed and heterogeneous systems
Moving from monolithic towards component-based approaches
Moving from defining the flow of execution towards definingcoordination among the components
Interaction protocols
are means for achieving the desired coordination among the components
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 2 / 41
Interaction Protocols
Representation of Complex Systems
Current trends:
Advent of distributed and heterogeneous systems
Moving from monolithic towards component-based approaches
Moving from defining the flow of execution towards definingcoordination among the components
Interaction protocols
are means for achieving the desired coordination among the components
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 2 / 41
Interaction Protocols
Interaction Protocols
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 3 / 41
Achieve coordination
Providing all the necessary tools
Accounting for components’ characteristics
Interaction Protocols
Interaction Protocols
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 3 / 41
Achieve coordination
Providing all the necessary tools
Accounting for components’ characteristics
Reuse
Improving performances
Decreasing costs
Interaction Protocols
Interaction Protocols
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 3 / 41
Achieve coordination
Providing all the necessary tools
Accounting for components’ characteristics
Reuse
Improving performances
Decreasing costs
Applicable to real cases
Interaction Protocols
Interaction Protocols
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 3 / 41
Achieve coordination
Providing all the necessary tools
Accounting for components’ characteristics
Reuse
Improving performances
Decreasing costs
Usable
Tools
Methodologies
Supports
Applicable to real cases
Interaction Protocols
Interaction Protocols
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 3 / 41
Achieve coordination
Providing all the necessary tools
Accounting for components’ characteristics
Formal Foundation
Reuse
Improving performances
Decreasing costs
Usable
Tools
Methodologies
Supports
Applicable to real cases
Interaction Protocols
Interaction Protocol and MASs
A MAS is an abstraction used to simulate complex systems made ofinteracting parties (agents) where:
the interacting parties are usually conceived as heterogeneous andautonomous;they interact in order to achieve some objectives.
Interaction protocols are means to achieve coordination/cooperationin MASs [Weiss, 1999, Wooldridge, 2002]
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 4 / 41
Interaction Protocols
How to specify an Interaction Protocol?
By specifying the allowed sequences of actions
Automata
Finite State Machines
Petri Nets
UML Diagrams (Sequence diagrams)
AUML Diagrams (Interaction diagrams)
By specifying only mandatory and forbidden requirements
Expectation-based approaches
Commitment-based approaches
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 5 / 41
Interaction Protocols
How to specify an Interaction Protocol?
By specifying the allowed sequences of actions
Automata
Finite State Machines
Petri Nets
UML Diagrams (Sequence diagrams)
AUML Diagrams (Interaction diagrams)
By specifying only mandatory and forbidden requirements
Expectation-based approaches
Commitment-based approaches
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 5 / 41
Interaction Protocols Commitment Protocols
Commitment Protocols
C(x , y , p, q)
In Commitment protocols [Yolum and Singh, 2002]
By performing some actions the agents socially commit to do (orachieve) something
The expectation is that at the end all commitments are satisfied
Characteristics
they respect the agents’ autonomy
they allow agents for flexible behaviours (what rather than how)
they are verifiable
However,
Commitment Protocols do not account for Patterns of Interaction
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 6 / 41
Interaction Protocols Commitment Protocols
Commitment Protocols
C(x , y , p, q)
In Commitment protocols [Yolum and Singh, 2002]
By performing some actions the agents socially commit to do (orachieve) something
The expectation is that at the end all commitments are satisfied
Characteristics
they respect the agents’ autonomy
they allow agents for flexible behaviours (what rather than how)
they are verifiable
However,
Commitment Protocols do not account for Patterns of Interaction
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 6 / 41
Interaction Protocols Commitment Protocols
Commitment Protocols
C(x , y , p, q)
In Commitment protocols [Yolum and Singh, 2002]
By performing some actions the agents socially commit to do (orachieve) something
The expectation is that at the end all commitments are satisfied
Characteristics
they respect the agents’ autonomy
they allow agents for flexible behaviours (what rather than how)
they are verifiable
However,
Commitment Protocols do not account for Patterns of Interaction
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 6 / 41
Interaction Protocols Commitment Protocols
Commitment Protocols
C(x , y , p, q)
In Commitment protocols [Yolum and Singh, 2002]
By performing some actions the agents socially commit to do (orachieve) something
The expectation is that at the end all commitments are satisfied
Characteristics
they respect the agents’ autonomy
they allow agents for flexible behaviours (what rather than how)
they are verifiable
However,
Commitment Protocols do not account for Patterns of Interaction
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 6 / 41
Interaction Protocols Patterns of Interaction
Patterns of interaction
They can be used for expressing requirements on the evolution of theinteraction (e.g. agreements, norms, rules, conventions...) improvingcoordination
They bring to the same state but...
one may wish to say that only payment followed by shipment is acceptable
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 7 / 41
Main Contributions
Main Contributions
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 8 / 41
Interaction and coordina-tion among agents
Main Contributions
2CL Protocols
Extend commitment protocols with the possibility of expressing patterns ofinteraction
The specification is based on Constraints and Commitments:
Commitments: Capture contractual/social relations among the interactingparties
Constraints: Are used to represent patterns of interaction that theinteraction is desired to respect.
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 9 / 41
· M. Baldoni, C. Baroglio, E. Marengo, and V. Patti. Constitutive and Regulative Specificationsof Commitment Protocols: a Decoupled Approach. ACM Transactions on Intelligent Systemsand Technology, Special Issue on Agent Communication, To appear.· M. Baldoni, C. Baroglio, and E. Marengo. Behavior-oriented Commitment- based Protocols. InProc. of ECAI 2010.· M. Baldoni, C. Baroglio, and E. Marengo. Constraints among Commitments: Regulative Spec-ification of Interaction Protocols. In Proc. of AC 2010.
Main Contributions
Main Contributions
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 10 / 41
Interaction and coordina-tion among agents
TheoreticalFramework
Main Contributions
Main Contributions
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 10 / 41
Interaction and coordina-tion among agents
TheoreticalFramework
Evaluation
Main Contributions
Main Contributions
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 10 / 41
Interaction and coordina-tion among agents
TheoreticalFramework
Adaptation/Grafting
Evaluation
Main Contributions
Main Contributions
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 10 / 41
Interaction and coordina-tion among agents
TheoreticalFramework
Adaptation/Grafting
Protocol Engineering,Design and Analysis
Evaluation
2CL Protocols Specification
Interaction among Agents
A 2CL interaction protocol P is a tuple
Ro: interacting roles
F: set of literals (facts and commitments)
s0: initial state
A: set of protocol actions
Cst: set of constraints defining the patterns of interaction
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 11 / 41
2CL Protocols Specification Constitutive Specification
Constitutive Specification
X means Y if Cond
The set of actions A, defined on Ro and F, forms the ConstitutiveSpecification [Searle, 1995] of a 2CL protocol
Count-as relation
X is a physical event, Y is a (set of) social event(s), Cond is the conditionunder which X acquires the specified meaning
means captures a count-as relation between physical and socialevents
social events correspond to operations on facts and commitments
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 12 / 41
2CL Protocols Specification Constitutive Specification
Constitutive Specification
X means Y if Cond
The set of actions A, defined on Ro and F, forms the ConstitutiveSpecification [Searle, 1995] of a 2CL protocol
Count-as relation
X is a physical event, Y is a (set of) social event(s), Cond is the conditionunder which X acquires the specified meaning
means captures a count-as relation between physical and socialevents
social events correspond to operations on facts and commitments
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 12 / 41
2CL Protocols Specification Constitutive Specification
Example
Constitutive Specification of CNET
(a) cfp means create(C(i , p, proposal , assigned ∨ send reject))
(b) proposal means create(C(p, i , assigned , solved)) if ¬solved
(c) send refusal means refused task if cfp ∧ ¬solved
(d) send accept means assigned
(e) send reject means release(C(p, i , assigned , solved)) if proposal
(f) send done means solved
(g) failure means cancel(C(p, i , assigned , solved))∧ cancel(C(p, i , solved))
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 13 / 41
2CL Protocols Specification Constraints Specification
Constraints Specification
dnf1 op dnf2
The set of constraints Cst, defined on Ro and on F, extends the regulativespecification [Searle, 1995] of a commitment protocol.
2CL: the Constraints among Commitments Language
Allows to express patterns as constraints among facts and commitments
dnf1 and dnf2 are disjunctive normal form formulas; op is a 2CLconstraint
It is a declarative language
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 14 / 41
2CL Protocols Specification Constraints Specification
Constraints Specification
dnf1 op dnf2
The set of constraints Cst, defined on Ro and on F, extends the regulativespecification [Searle, 1995] of a commitment protocol.
2CL: the Constraints among Commitments Language
Allows to express patterns as constraints among facts and commitments
dnf1 and dnf2 are disjunctive normal form formulas; op is a 2CLconstraint
It is a declarative language
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 14 / 41
2CL Protocols Specification 2CL
The Constraints among Commitments Language
Operator Meaning
RelationOperators
dnf1 correlate dnf2 In an execution where dnf1 occurs, also dnf2 must occurbut there is no temporal relation between the two.
dnf1 not correlate dnf2 If dnf1 occurs in some execution, dnf2 must not occur.
dnf1 co-exist dnf2 It captures the mutual correlation dnf1 correlate dnf2 anddnf2 correlate dnf1.
dnf1 not co-exist dnf2 This captures the mutual exclusion of dnf1 and dnf2:both dnf1 not correlate dnf2 and dnf2 not correlate dnf1hold.
TemporalOperators
dnf1 response dnf2 If dnf1 occurs, dnf2 must hold at least once afterwards(or in the same state). It does not matter if dnf2 alreadyheld before dnf1.
dnf1 not response dnf2 If dnf1 holds, dnf2 cannot hold in the same state or after.
dnf1 before dnf2 dnf2 cannot hold until dnf1 becomes true. Afterwards, itis not necessary that dnf2 becomes true.
dnf1 not before dnf2 In case dnf2 becomes true, dnf1 cannot hold beforehand.
dnf1 cause dnf2 It is the conjunction of response and before relations: dnf1response dnf2 and dnf1 before dnf2.
dnf1 not cause dnf2 It is the conjunction of response and before negative rela-tions: dnf1 not response dnf2 and dnf1 not before dnf2.
StrongSequence
dnf1 premise dnf2 dnf1 must hold in the state immediately preceding onestate in which dnf2 holds.
dnf1 not premise dnf2 dnf1 must never hold in a state that immediately precedesone where dnf2 holds.
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 15 / 41
2CL Protocols Specification 2CL
Example
Constraints Specification of CNET
(c1) cfp before refused task or C(p, i , assigned , solved)
(c2) C(p, i , assigned , solved) before send reject or assigned
(c3) assigned before solved
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 16 / 41
2CL Protocols Specification 2CL
Example
Constraints Specification of CNET
(c1) cfp before refused task or C(p, i , assigned , solved)
(c2) C(p, i , assigned , solved) before send reject or assigned
(c3) assigned before solved
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 16 / 41
2CL Protocols Specification 2CL
Example
Constraints Specification of CNET
(c1) cfp before refused task or C(p, i , assigned , solved)
(c2) C(p, i , assigned , solved) before send reject or assigned
(c3) assigned before solved
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 16 / 41
2CL Protocols Specification 2CL
2CL Constraints Graphical representation
Declarative Approaches
are accused of being not very intuitive (lowering their usability)[Miller and McGinnis, 2008]
lack of graphical intuitive representations oriented to designers
Graphical convention
It is inspired by Declare [van der Aalst and Pesic, 2006, Montali, 2009]and DCML [Baldoni et al., 2007, Baldoni et al., 2011b] representations
dot represents the “triggering condition”
arrow represents the temporal nature of the constraint
negated operators are crossed by a line
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 17 / 41
· M. Baldoni, C. Baroglio, E. Marengo, and V. Patti. Constitutive and Regulative Specificationsof Commitment Protocols: a Decoupled Approach. ACM Transactions on Intelligent Systemsand Technology, Special Issue on Agent Communication, To appear.· M. Baldoni, C. Baroglio, and E. Marengo. Constraints among Commitments: Regulative Spec-ification of Interaction Protocols. In Proc. of AC 2010.
2CL Protocols Specification 2CL
2CL Constraints Graphical representation
Declarative Approaches
are accused of being not very intuitive (lowering their usability)[Miller and McGinnis, 2008]
lack of graphical intuitive representations oriented to designers
Graphical convention
It is inspired by Declare [van der Aalst and Pesic, 2006, Montali, 2009]and DCML [Baldoni et al., 2007, Baldoni et al., 2011b] representations
dot represents the “triggering condition”
arrow represents the temporal nature of the constraint
negated operators are crossed by a line
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 17 / 41
· M. Baldoni, C. Baroglio, E. Marengo, and V. Patti. Constitutive and Regulative Specificationsof Commitment Protocols: a Decoupled Approach. ACM Transactions on Intelligent Systemsand Technology, Special Issue on Agent Communication, To appear.· M. Baldoni, C. Baroglio, and E. Marengo. Constraints among Commitments: Regulative Spec-ification of Interaction Protocols. In Proc. of AC 2010.
2CL Protocols Specification 2CL
Graphical representation: Relation Operators
Constraint Representation
Correlationl1 correlate l2 l1 l2
l1 not correlate l2 l1 l2
Co-existencel1 co-exist l2 l1 l2
l1 not co-exist l2 l1 l2
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 18 / 41
2CL Protocols Specification 2CL
Graphical representation: Temporal Operators
Constraint Representation
Responsel1 response l2 l1 l2
l1 not response l2 l1 l2
Beforel1 before l2 l1 l2
l1 not before l2 l1 l2
Causel1 cause l2 l1 l2
l1 not cause l2 l1 l2l1 l2
Premisel1 premise l2 l1 l2
l1 not premise l2 l1 l2
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 19 / 41
2CL Protocols Specification 2CL
Graphical representation: Drawing Formulas
Literal representation
l
Conjunction of literals
. . .l1 ln
cf = l1 ∧ . . . ∧ lnDisjunctive normal form formula
lnl1 . . .. . .cfi
dnf = l1 ∨ . . . ∨ cfi ∨ . . . ∨ ln
Exclusive Disjunctive normal form formula
lnl1 . . .. . .cfi
dnf = l1 xor . . . xor cfi xor . . . xor ln
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 20 / 41
2CL Protocols Specification 2CL
Graph of Constraints: no-flow-in-flow
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 21 / 41
2CL Protocols Specification 2CL
Graph of Constraints: no-flow-in-flow
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 22 / 41
2CL Protocols Specification Comparison with other Proposals
Comparison with other Proposals
Current Proposals
Do not account for patterns of interaction [Chopra and Singh, 2008, Chopra, 2009]
Focus on regulative aspects[Chesani et al., 2009, Montali et al., 2010, Torroni et al., 2009]
Patterns are hidden inside actions definitions[Winikoff et al., 2005, Chopra and Singh, 2006, Kafali and Yolum, 2009]
Represent patterns procedurally[Fornara and Colombetti, 2003, Fornara and Colombetti, 2004]
Capture some relations expressed among events[Mallya and Singh, 2006][Singh, 2003, Marengo et al., 2011]
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 23 / 41
M. Baldoni, C. Baroglio, and E. Marengo.Commitment-based Protocols with BehavioralRules and Correctness Properties of MAS. In Post-Proc. of DALT 2010, Revised Selected and InvitedPapers, LNAI 6619.
Main Contributions
Main Contributions
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 24 / 41
Interaction and coordina-tion among agents
Evaluation
Main Contributions
Main Contributions
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 24 / 41
Interaction and coordina-tion among agents
Evaluation
Adaptation/Grafting
Evaluation
Evaluation
Case Studies
OECD Guidelines for private data protection
MiFID: Markets in Financial Instruments Directive (investmentservices off-site)
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 25 / 41
·M. Baldoni, C. Baroglio, E. Marengo, and V. Patti. Grafting Regulations into BusinessProtocols: Supporting the Analysis of Risks of Violation. In Proc. of RELAW 2011.· M. Baldoni, C. Baroglio, E. Marengo, V. Patti, and F. Capuzzimati. Learn the Rulesso you Know How to Break them Properly. In Proc. of WOA 2011.· M. Baldoni, C. Baroglio, V. Patti, and E. Marengo. Supporting the Analysis of Risksof Violation in Business Protocols: the MiFID Case Study. In Information Systems:Crossroads for Organization, Management, Accounting and Engineering. Springer,2012. Best Track Paper Award.
Evaluation
Grafting of new Regulations
Data Flow Protocol
(a) ask data(b) send data(c) refuse data
OECD Guidelines
(d) periodically verify accuracy(e) check accuracy(f) verify purpose(g) notify owner
Constraints specifying the grafting
(c1) purpose verified −.• sent data(c2) accuracy verified −.• sent data(c3) sent data •−. owner notified(c4) purpose verified −.• refuse data(c5) accuracy verified −.• refuse data
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 26 / 41
Evaluation
Grafting of new Regulations
1
start
4
5
verify_purpose !c2
7
check_accuracy !c1
6
check_accuracy verify_purpose
12
13
check_accuracy
15
verify_purpose !c5
16
check_accuracy !c4
verify_purpose
3
notify_owner !c2
8
verify_purpose !c2
10
check_accuracy !c1
notify_owner !c2
9
check_accuracy notify_owner !c1verify_purpose
notify_owner
2
refuse_data !c4 send_data !c2
11
verify_purpose
17
check_accuracy
refuse_data !c5 send_data !c2
14
check_accuracy
send_datarefuse_data
refuse_data !c4 send_data !c1verify_purpose
ask_data
18
periodically_verify_accuracy
ask_data
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 27 / 41
·M. Baldoni, C. Baroglio, E. Marengo, and V. Patti. Grafting Regulations into BusinessProtocols: Supporting the Analysis of Risks of Violation. In Proc. of RELAW 2011.· M. Baldoni, C. Baroglio, V. Patti, and E. Marengo. Supporting the Analysis of Risksof Violation in Business Protocols: the MiFID Case Study. In Information Systems:Crossroads for Organization, Management, Accounting and Engineering. Springer,2012. Best Track Paper Award.
Evaluation MiFID
Grafting of new Regulations
Sale Protocol
(a) propose solution(b) accept proposal(c) introduce investor(d) prepare contract(e) sign contract(f) countersign contract(g) notify(h) invest(i) withdraw
Introducing MiFID
(j) interview(k) profile(l) classify products(m) evaluate(n) verify
Constraints specifying the Grafting
(c1) document supplied −.• C(bank, inv , invest)(c2) document supplied −.• C(inv , bank, prepare contract, sign contract)
∧C(inv , bank, sign contract)(c3) interview −.• introduce investor(c4) verify −.• countersign contract(c5) profile −.• C(fp, inv , proposed riskL)(c6) evaluate −.• proposed riskL
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 28 / 41
Evaluation MiFID
Labelled Graph after MiFID
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 29 / 41
Main Contributions
Main Contributions
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 30 / 41
Interaction and coordina-tion among agents Adaptation/Grafting
Protocol Engineering,Design and Analysis
Evaluation
Support to usability Methodology
2CM: the Constraints among CommitmentsMethodology
2CM supports the analyst in:
Protocol Specification
Protocol Composition
Protocol Specialization
It adapts the Amoeba [Desai et al., 2009] methodology to handling 2CLprotocols and extends it for protocol Specialization
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 31 / 41
Support to usability Tools
Tool
It is realized as an Eclipse plug-in:
Translating specification files
Support for protocol specification
Generation and Visualization of the Labelled Graph of PossibleInteractions
Exploration of the labelled graph:
one node at a time;legal paths;illegal paths;content of states.
http://di.unito.it/2cl
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 32 / 41
M. Baldoni, C. Baroglio, F. Capuzzimati, E. Marengo, and V.Patti. A Generalized Commitment Machine for 2CL Protocolsand its Implementation. Proc. of DALT 2012.
Support to usability Tools
Labelled Graph of Interactions
Aims at
supporting the analysis of the specification (e.g. the Analysis of risks ofviolations)
It is realized in tuProlog
It extends the enhanced commitment machine by Winikoff et al.[Winikoff et al., 2005, DALT]
Enhanced Commitment Machine features:
Generation of reachable statesCommitments life-cycle
Additionally we evaluate protocol constraints
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 33 / 41
Main Contributions
Main Contributions
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 34 / 41
Interaction and coordina-tion among agents
TheoreticalFramework
Adaptation/Grafting
Protocol Engineering,Design and Analysis
Evaluation
2CL-GCM
Theoretical Framework
2CL-GCM: 2CL-Generalized Commitment Machine
2CL-GCM extends the Generalized Commitment Machine (GCM) bySingh [Singh, 2007, IJCAI]
GCM allows to infer transitions among the states
Given a set of statesGiven a set of actions
2CL-GCM extends GCM
accounting for a set of constraintsconstraints are used to characterise a path of the 2CL-GCM
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 35 / 41
M. Baldoni, C. Baroglio, F. Capuzzimati, E. Marengo, and V.Patti. A Generalized Commitment Machine for 2CL Protocolsand its Implementation. Proc. of DALT 2012.
2CL-GCM
LTL interpretation of 2CL constraints
Protocol constraints Cst
To be a path of a 2CL-GCM a sequence of states must satisfy theconstraints in Cst
To this aim we associate an LTL interpretation (ϕltl ) to each 2CLoperator
Relation Operator LTL interpretation ϕltl
RelationOperators
CorrelationA correlate B ♦A→ ♦B
A not correlate B ♦A→ ¬♦B
Co-existenceA co-exist B ϕltl (A correlate B) ∧ ϕltl (B correlate A)
A not co-exist B ϕltl (A not correlate B) ∧ ϕltl (B not correlate A)
TemporalOperators
ResponseA response B �(A→ ♦B)
A not response B �(A→ ¬♦B)
BeforeA before B ¬B ∪ A
A not before B �(♦B → ¬A)
CauseA cause B ϕltl (A response B) ∧ ϕltl (A before B)
A not cause B ϕltl (A not response B) ∧ ϕltl (A not before B)
StrongSequence
PremiseA premise B �(©B → A)A not premiseB �(©B → ¬A)
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 36 / 41
M. Baldoni, C. Baroglio, and E. Marengo.Behavior-oriented Commitment- basedProtocols. In Proc. of ECAI 2010.
2CL-GCM
2CL-Generalized Commitment Machine (2CL-GCM)
2CL-GCM
A 2CL-GCM is a tuple P = 〈S ,A, s0,∆,G ,Cst〉, where:
- S is a finite set of states;
- A is a finite set of actions;
- s0 ∈ S is the initial state;
- ∆ is an action theory;
- G ⊆ S is a set of good states;
- Cst is a set of 2CL constraints.
(i) Members of S are logically distinct, that is: ∀s, s ′ ∈ S , s 6≡ s ′; (ii)false 6∈ S ; and (iii) any state that logically derives a good state is alsogood.
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 37 / 41
2CL-GCM
Path of a 2CL-GCM
Path of a 2CL-GCM
Is a path according to the GCM and
It satisfies the LTL interpretation associated to each 2CL constraintof the specification
2CL-GCM path
Let P = 〈S ,A, s0,∆,G ,Cst〉 be a 2CL-GCM. Let τ = 〈(τ0, a0, τ1), . . . 〉be an infinite sequence of triples and T (τ) be the corresponding transitionsystem. Let inf (τ) be the set of states that occur infinitely often in τ . τ isa path generated from P when:
(i) ∀(τi , ai , τi+1) in τ then τi , τi+1 ∈ S and ai ∈ A and
τiai↪→ τi+1 ∈ ∆; and
(ii) inf (τ) ∩ G 6= ∅; and
(iii) ∀c ∈ Cst : T (τ), τ0 |=LTL ϕltl (c)
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 38 / 41
Conclusions
Conclusions and Future Work
2CL Protocols
Allow for the specification of patterns of interaction in protocolspecifications
The specification is flexible
They foster reuse of the specification by supporting the adaptation ofa specification to different contexts
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 39 / 41
Conclusions
Conclusions and Future Work
Ongoing and Future Directions
Introduce 2CL constraints as commitment conditions:
along the line of Regula [Marengo et al., 2011];explicitly identifying a responsible;modifying commitments’ life cycle;notions of safety and control.
Study new functionalities for the tool and explore the adoption ofdifferent languages
Exploit the A&Ameta-model [Baldoni et al., 2010, Baldoni et al., 2011a]:
capturing direct and indirect communication;supporting observability [Baldoni and Baroglio, 2012];monitoring the interaction.
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 40 / 41
Conclusions
Thanks
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 41 / 41
Conclusions
Baldoni, M. and Baroglio, C. (2012).Some Thoughts about Commitment Protocols.In Baldoni, M., Dennis, L., Mascardi, V., and Vasconcelos, W.,editors, Proc. of International Workshop on Declarative AgentLanguages and Technologies, DALT 2012, held in conjuction withAAMAS 2012, Valencia, Spain.
Baldoni, M., Baroglio, C., Bergenti, F., Boccalatte, A., Marengo, E.,Martelli, M., Mascardi, V., Padovani, L., Patti, V., Ricci, A., Rossi,G., and Santi, A. (2010).MERCURIO: An Interaction-oriented Framework for Designing,Verifying and Programming Multi-Agent Systems.In Fornara, N. and Vouros, G., editors, Proc. of the 3rd Multi-AgentLogics, Languages, and Organisations Federated Workshops(MALLOW’10), 11th International Workshop on Coordination,Organization, Institutions and Norms in Multi-Agent Systems(COIN@MALLOW 2010), volume 627, Domain Valpre in Lyon,France. CEUR Workshop Proceedings.
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 41 / 41
Conclusions
Baldoni, M., Baroglio, C., Bergenti, F., Marengo, E., Mascardi, V.,Patti, V., Ricci, A., and Santi, A. (2011a).An Interaction-oriented Agent Framework for Open Environments.In Pirrone, R. and Sorbello, F., editors, Proc. Artificial IntelligenceAround Man and Beyond, 12th Congress of the Italian Association forArtificial Intelligence (AI*IA 2011), volume 6934 of LNAI, pages68–79, Palermo, Italy. Springer.
Baldoni, M., Baroglio, C., Brunkhorst, I., Henze, N., Marengo, E., andPatti, V. (2011b).Constraint Modeling for Curriculum Planning and Validation.International Journal of Interactive Learning Environments,19(1):83–123.
Baldoni, M., Baroglio, C., and Marengo, E. (2007).Curricula Modeling and Checking.In Basili, R. and Pazienza, M. T., editors, Proc. of AI*IA 2007:Advances in Artificial Intelligence, 10th Congress of the Italian
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 41 / 41
Conclusions
Association for Artificial Intelligence, volume 4733 of LNAI, pages471–482, Rome, Italy. Springer.
Chesani, F., Mello, P., Montali, M., and Torroni, P. (2009).Commitment Tracking via the Reactive Event Calculus.In Boutilier, C., editor, IJCAI, pages 91–96, Pasadena, California, USA.
Chopra, A. (2009).Commitment Alignment: Semantics, Patterns, and DecisionProcedures for Distributed Computing.PhD thesis, North Carolina State University, Raleigh, NC.
Chopra, A. K. and Singh, M. P. (2006).Contextualizing Commitment Protocol.In Nakashima, H., Wellman, M. P., Weiss, G., and Stone, P., editors,Proc. of 5th International Joint Conference on Autonomous Agentsand Multiagent Systems (AAMAS 2006), pages 1345–1352,Hakodate, Japan. ACM.
Chopra, A. K. and Singh, M. P. (2008).
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 41 / 41
Conclusions
Constitutive Interoperability.In Proc. of the 7th international joint conference on Autonomousagents and multiagent systems, volume 2, pages 797–804.International Foundation for Autonomous Agents and MultiagentSystems.
Desai, N., Chopra, A. K., and Singh, M. P. (2009).Amoeba: A Methodology for Modelling and EvolvingCross-Organizational Business Processes.ACM Transactions on Software Engineering and Methodology, 19(2).
Fornara, N. and Colombetti, M. (2003).Defining Interaction Protocols using a Commitment-based AgentCommunication Language.In [Rosenschein et al., 2003], pages 520–527.
Fornara, N. and Colombetti, M. (2004).A Commitment-Based Approach To Agent Communication.Applied Artificial Intelligence, 18(9-10):853–866.
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 41 / 41
Conclusions
Kafali, O. and Yolum, P. (2009).Detecting Exceptions in Commitment Protocols: Discovering HiddenStates.In Proc. of the Second Multi-Agent Logics, Languages, andOrganisations Federated Workshops (MALLOW 2009), volume 494 ofCEUR Workshop Proceedings, Turin, Italy. CEUR-WS.org.
Mallya, A. U. and Singh, M. P. (2006).Introducing Preferences into Commitment Protocols.In Dignum, F., van Eijk, R. M., and Flores, R. A., editors, AgentCommunication II, International Workshops on Agent Communication(AC 2005 and AC 2006), volume 3859 of LNCS, pages 136–149,Utrecht, Netherlands. Springer.
Marengo, E., Baldoni, M., Baroglio, C., Chopra, A. K., Patti, V., andSingh, M. P. (2011).Commitments with Regulations: Reasoning about Safety and Controlin REGULA.
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 41 / 41
Conclusions
In Sonenberg, L., Stone, P., Tumer, K., and Yolum, P., editors,AAMAS, volume 1–3, pages 467–474, Taipei, Taiwan. IFAAMAS.
Miller, T. and McGinnis, J. (2008).Amongst First-Class Protocols.In Proc. of Eng. Societies in the Agents World VIII, volume 4995 ofLNCS, pages 208–223. Springer.
Montali, M. (2009).Specification and Verification of Declarative Open Interaction Models.PhD thesis, Electronics, Computer Science and TelecommunicationsEngineering, University of Bologna.
Montali, M., Torroni, P., Chesani, F., Mello, P., Alberti, M., andLamma, E. (2010).Abductive Logic Programming as an Effective Technology for theStatic Verification of Declarative Business Processes.Fundamenta Informaticae, 102(3-4):325–361.
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 41 / 41
Conclusions
Rosenschein, J. S., Sandholm, T., Wooldridge, M., and Yokoo, M.,editors (2003).Proc. of the Second International Joint Conference on AutonomousAgents & Multiagent Systems (AAMAS 2003), Melbourne, Australia.ACM.
Searle, J. R. (1995).The construction of social reality.Free Press, New York.
Singh, M. P. (2003).Distributed Enactment of Multiagent Workflows: Temporal Logic forWeb Service Composition.In [Rosenschein et al., 2003], pages 907–914.
Singh, M. P. (2007).Formalizing Communication Protocols for Multiagent Systems.In Veloso, M. M., editor, IJCAI, pages 1519–1524, Hyderabad, India.
Torroni, P., Chesani, F., Mello, P., and Montali, M. (2009).
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 41 / 41
Conclusions
Social Commitments in Time: Satisfied or Compensated.In Baldoni, M., Bentahar, J., van Riemsdijk, M. B., and Lloyd, J.,editors, DALT, volume 5948 of LNCS, pages 228–243, Budapest,Hungary. Springer.
van der Aalst, W. M. P. and Pesic, M. (2006).DecSerFlow: Towards a Truly Declarative Service Flow Language.In Leymann, F., Reisig, W., Thatte, S. R., and van der Aalst, W.M. P., editors, The Role of Business Processes in Service OrientedArchitectures, volume 06291 of Dagstuhl Seminar Proceedings.Internationales Begegnungs- und Forschungszentrum fuer Informatik(IBFI), Schloss Dagstuhl, Germany.
Weiss, G., editor (1999).Multiagent Systems: A Modern Approach to Distributed ArtificialIntelligence.The MIT Press.
Winikoff, M., Liu, W., and Harland, J. (2005).Enhancing Commitment Machines.
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 41 / 41
Conclusions
In DALT 2004, volume 3476 of LNCS, pages 198–220. Springer.
Wooldridge, M. (2002).An Introduction to Multiagent Systems.John Wiley & Sons.
Yolum, P. and Singh, M. P. (2002).Commitment Machines.In Meyer, J.-J. C. and Tambe, M., editors, Proc. of the 8thInternational Workshop on Intelligent Agents VIII (ATAL 2001),volume 2333 of LNCS, pages 235–247, Seattle, WA, USA. Springer.
Elisa Marengo (UNITO) 2CL Protocols October 19, 2012 41 / 41
Top Related