Programming Open Systems with Agents, Environments and Organizations

Transcription

1 Programming Open Systems with Agents, Environments and s: Michele Piunti 1 Ricci 1 Olivier Boissier 2 Jomi 3 1 Università degli studi di Bologna - DEIS, Bologna - Italy. {michele.piunti a.ricci}@unibo.it 2 Ecole Nationale Superieure des Mines - G2I St-Etienne - France boissier@emse.fr 3 Federal University of Santa Catarina - DAS, Florianópolis - Brazil jomi@das.ufsc.br WOA Rimini

2 s: 1 2 s: Outline 3 4

3 Current Issues in MAS Programming s: BELIEFS GOALS PLANS PERCEPTIONS AGENTS INTERNAL EVENTS ROLES MISSIONS DUTIES ACTIONS? DEONTIC RELATIONS NORMS SANCTIONS/REWARDS ORGANIZATIONS RESOURCES LEGACY SERVICES COMPONENTS ENVIRONMENTS Seamless integration of multiple entities and mechanisms Interaction, Openness

4 Current Approaches in MAS Programming s: AGENT(S) A-E A-O ORGANIZATION O-E ENVIRONMENT

5 Current Approaches in MAS Programming s: AGENT(S) Jason [Bordini et al., 2007], Jadex [Pokahr et al., 2005], 2APL [Dastani, 2008], etc. A-E A-O ORGANIZATION AGR/MADKIT [Ferber et al., 2003], 2OPL [Baldoni et al., 2008], Electronic Institutions [Esteva et al., 2004], S-MOISE + [Hübner et al., 2005], OPERA [Dignum, 2003], etc. O-E ENVIRONMENT CArtAgO [Ricci et al., 2010], GOLEM [Bromuri and Stathis, 2007], MASQ, AGRE [Stratulat et al., 2009, Báez-Barranco et al., 2006], Normative Objects [Okuyama et al., 2009], Situated Electronic Institutions [Campos et al., 2008], Brahms [Sierhuis, 2001] etc.

6 A-E-O Integrations s: AGENT(S) A-E A-O ORGANIZATION O-E ENVIRONMENT

7 s: The Idea: s Human s Analogy instrumentation of the environment with specific Infrastructures that are explicitely conceived for easing humans activities/tasks in organisations [Kirsh, 1995]. Environment Infrastructures Artifacts Hospital Scenario Example: A visitor entering the hospital has not a complete knowledge of the organisation behind Visitors (as users) ignore the mechanisms and the structures at the basis of the organisation Patient Agents Nonetheless, visitors find those things which they are interested in, means to achieve their goals finally Staff Agent Visitor Agents Staff Agents

8 s: Expected Outcomes Integration and Synergy between E and O enabling agents to profitably interact with both organisation and other environmental entities; enabling organisation to control agents and regiment environmental resources; allowing environmental changes to affect both organisation dynamics and agents activities; At the application level, possibility: To reconcile agents and their work environments with institutional dimensions (i.e. organisations); To exploit a strong notion of agency, i.e., mental attitudes (purposes, knowledge), events, perception To cope with Interoperability and Openness

9 s: 1 2 s: Outline 3 4

10 s: Constitutive Rules [Searle, 1964] Constitutive Rules Typical of human societies (Social Reality [Searle, 1997] ) The reification of a state in a particular context may constitute the realization of a particular institutional/organizational fact AGENT(S) A-O A-E Enact ENVIRONMENT Count-As ORGANIZATION

11 s: Agents Emb-Org-Rules Organization Rules Environment Management Infrastructure count-as enact count-as Management Infrastructure Used to automate particular dynamics between E-O:

12 s: Agents Emb-Org-Rules Organization Rules Environment Management Infrastructure count-as enact count-as Management Infrastructure Used to automate particular dynamics between E-O: Entering an ambulatory room count-as adopting the role patient Finalizing the payment operation on the billing machine count-as achieving the goal pay A sold out in the visit schedule enact the suspension of the booking service

13 Meta- s: Constitutive Rule (Emb-Org-Rule) 1..n Organization Count-as Rule Enact Rule

14 Meta- s: Ev Type Ev Value Event Triggers Constitutive Rule (Emb-Org-Rule) 1..n Environment Event Organization Event Organization Produces Count-as Rule Enact Rule

15 s: Rules Event-Condition-Action (ECA) rules: when ev in the context c apply oeexp +ev : c oeexp ev are organisational event (org-ev) or environment event (env-ev) c refers to observable states of the or of the Environment oeexp refers to organisation embodiement expressions

16 s: Rules Event-Condition-Action (ECA) rules: when ev in the context c apply oeexp +ev : c oeexp ev are organisational event (org-ev) or environment event (env-ev) c refers to observable states of the or of the Environment oeexp refers to organisation embodiement expressions Embodiement Expressions include workspace operators: (1) applyop(target id,op name [,Params]) (2) applylop(target id,op name [,Params]) (3) disable(target id [,ag id ] {,op name }) (4) enable(target id [,ag id ] {,op name }) (5) make(target id,target n [,Params]) (6) dispose(target id ) (7) exclude(ag id ) (8) include(ag id ) where target id is the management component id, op name is an organisation operation (org-op) or environment operation (env-op)

17 s: 1 2 s: Outline 3 4

18 A Concrete s: STAFF VISITOR Terminal BillingMachine EMI ENVIRONMENT ARTIFACTS GroupBoards OMI SchemeBoards ORGANISATIONAL ARTIFACTS Hospital Workspace Agent s SurgeryTablet Desk VISITOR STAFF EMI: Environment Management Infrastructure OMI: Management Infrastructure EMI based on CArtAgO [Piunti et al., 2008, Ricci et al., 2009]. env-op: operations provided by the physical artifacts env-ev: change of observable properties, results of operations,... OMI based on ORA4MAS [Hübner et al., 2009]: org-op: adopt-role, commit-mission, achieve-goal,... org-ev: any change of the organisation state

19 visitorsch visitor scheme Hospital Scenario monitorsch monitor scheme s: mvisit enter the room docsch Doctor scheme mdoc visit patient mvisit book the visit mpatient do the visit visit mpay pay visit ORGANIZATION MANAGEMENT INFRASTRUCTURE mstaff mvisit exit observe Hospital Workspace committed_agents goals_states scheme_specification obligations_state commitmission leavemission setgoalachieved Scheme artifacts mrew send bill enforcement player_agents schemes group_specification adoptrole leaverole addscheme msan send fee removescheme Group artifacts

20 visitorsch visitor scheme Hospital Scenario monitorsch monitor scheme s: mvisit enter the room joinworkspace Ambulatory Room docsch Doctor scheme mvisit book the visit use Desk bookvisit mpatient do the visit visit use SurgeryTablet signpat mpay pay visit use BillingMachine pay mvisit exit quitworkspace Ambulatory Room Ambulatory Room Workspace mstaff observe mrew send bill focus Desk, BillingMachine use Terminal sendbill reservations bookvisit enforcement visits signdoc signpat msan send fee use Terminal sendfee mdoc visit patient Desk SurgeryTablet payments sendbill use pay sendfee SurgeryTablet signdoc ENVIRONMENT MANAGEMENT BillingMachine Terminal INFRASTRUCTURE

21 Hospital Scenario Count-as Rule Example s: An event occurring in the system may count-as an institutional event and automatically update the organisation +join_req(ag) : true -> make("visitorgroupboard", "OMI.GroupBoard", ["moise/hospital.xml","visitgroup"]); make("visitorschboard", "OMI.SchemeBoard", ["moise/hospital.xml","visitorsch"]); apply("visitorgroupboard", adoptrole(ag, "patient")); include(ag). +op_completed("visitorgroupboard", _, adoptrole(ag, "patient")) : true -> apply("visitorschboard", commitmission(ag, "mpat")). +ws_leaved(ag) : true -> apply("visitorgroupboard", leaverole(ag, "patient")). +op_completed("billingmachine", Ag, pay) : true -> apply("visitorschboard", setgoalachieved(ag, pay_visit)). +op_completed("terminal", Ag, sendfee) : true -> apply("monitorschboard", setgoalachieved(ag, send_fee)).

22 Hospital Scenario Enact Rule Example s: may produce a control by enacting changes upon the environment (i.e., to promote equilibrium, avoid undesiderable states). +signal("visitorgroupboard", role_cardinality, visitor) : true -> disable("desk", bookvisit). +signal("monitorschboard", goal_non_compliance, obligation(ag, ngoa(monitorsch,mrew,send_bill), achieved(monitorsch,send_bill,ag), TTF) : true -> exclude(ag).

23 s: 1 2 s: Outline 3 4

24 s: An unifying approach to MAS programming ; s No need for agents to bring about organisational notions; Environment infrastructures succeed to mediate between agents and organisations; Global dynamics shaped on workspace events and transparently handled by the system. Limitations and Aspects we do not address (yet): Complex interaction patterns may result in many relationship to be specified between E-O. Direct communication between agents (Agent-Agent interaction) through message passing (i.e. ACL) is not currently under the control of the organisation.

25 s: Ongoing and Future Work: Perspectives To generalize the mechanism of Workspace Laws and Rules defining a wide set of inter-system functional relations (i.e. access control, security); To provide a general framweork for integrated MAS development Applications in future ICT: Any scenario integrating artificial agents, devices, humans in the same application Future Internet, Cloud Computing Sociotechnical systems, pervasive computing Virtualization, Electronic Marketplaces, etc.

26 s: Bibliography I Hübner, J. F., Boissier, O., Kitio, R., and Ricci, A. (2009). Instrumenting Multi-Agent s with al Artifacts and Agents. Journal of Autonomous Agents and Multi-Agent Systems. Kirsh, D. (1995). The intelligent use of space. Artificial Intelligence, 73(1-2): Piunti, M., Ricci, A., Braubach, L., and Pokahr, A. (2008). Goal-directed Interactions in Artifact-Based MAS: Jadex Agents playing in CArtAgO Environments. In IEEE/WIC/ACM International Joint Conference on Web Intelligence and Intelligent Agent Technology (WI-IAT 2008), Sydney, NSW, Australia. Ricci, A., Piunti, M., Viroli, M., and Omicini, A. (2009). Environment programming in CArtAgO. In Multi-Agent Programming: Languages, s and Applications, Vol. 2, pages Springer. Searle, J. R. (1964). Speech Acts, chapter What is a Speech Act? Cambridge University Press.

27 Bibliography II s: Searle, J. R. (1997). The Construction of Social Reality. Free Press. Stratulat, T., Ferber, J., and Tranier, J. (2009). MASQ: Towards an Integral Approach of Agent-Based Interaction. In Proc. of 8th Conf. on Agents and Multi Agent Systems (AAMAS-09). Tinnemeier, N., Dastani, M., Meyer, J.-J., and van der Torre, L. (2009). Programming normative artifacts with declarative obligations and prohibitions. In IEEE/WIC/ACM International Joint Conference on Web Intelligence and Intelligent Agent Technology (WI-IAT 2009).

28 Programming Open Systems with Agents, Environments and s: Michele Piunti 1 Ricci 1 Olivier Boissier 2 Jomi 3 1 Università degli studi di Bologna - DEIS, Bologna - Italy. {michele.piunti a.ricci}@unibo.it 2 Ecole Nationale Superieure des Mines - G2I St-Etienne - France boissier@emse.fr 3 Federal University of Santa Catarina - DAS, Florianópolis - Brazil jomi@das.ufsc.br WOA Rimini