motivation-based selection of negotiation opponents steve munroe and michael luck iam group...
Post on 01-Jan-2016
Embed Size (px)
Motivation-Based Selection of Negotiation OpponentsSteve Munroe and Michael LuckIAM GroupUniversity of Southampton
Presentation OutlineThe problem: dynamic domains and negotiationMotivated agents for dynamic domainsNegotiation goalsSelecting opponents
Autonomous NegotiationThree phases of negotiation
Dealing with dynamics and uncertainty
Strategies / Tactics / Protocols
Where do the issues come from?
How are reservation values determined?
Who best to negotiate with?
Autonomy and MotivationAutonomy:
The ability to make decisions and select courses of action that further ones own interests based on ones own assessment of the situation
Any desire or preference that can lead to the generation and adoption of goals and that affects the outcome of the reasoning or behavioural task intended to satisfy those goals
Motivation >>>> Utility
A Motivated Agent ArchitectureMotivational Cues
The Warehouse DomainController agent must negotiate with delivery agent about moving boxes around the warehouse
Negotiation GoalsDynamic reconfiguration of issues to meet current demandsFixed attributesPotential attribute
Negotiation Goals - 2Two types of potential attributesNon resource-basedResource-based
Non Resource-Based Attributes: constructing preferencesWhat is the structure of the preference?Determined by assessing each possibility in terms of motivational worth
Non Resource-Based Attributes: generating preferencesCalculates the worth of each of the possible values that can be used to instantiate a potential attribute
: v x gs x ms [0,1]
Non Resource-Based Attributes: attribute classification rulesFixed if the preferences of the agent contains at most one value that has positive motivational worth and all the rest have negative motivational worth.
Negotiable if the preferences of the agent contains more than one value that has positive motivational worth.
Slack if all the values contained in the agents preferences have the same motivational worth (both positive or negative).
Non Resource-Based Attributes: example preferences
Resource-Based Attributes: dynamic constraintsDynamic evaluation of resource useNever slack!Preference structure is monotonicProblem is to determine reservation on the use of a resource
Resource-Based Attributes: the reservation managerCalculate the unit worth of a resource, r, for an agent, a,
UWar = r bwar
where r = Imr [-1,1]
Then obtain the quantity of resource whose (negative) worth is equal to the worth of the goal.
Negotiation Goal StructureA negotiation goal contains sets ofFixed attributesNegotiable attributesSlack attributesReservation values for resource-dependent attributes
Selecting to minimise conflict
Selecting to optimise resource use
Selecting to Minimise ConflictEach agent selects its own negotiable attributesIntersection of choices defines issuesSmaller intersections means less conflict (easier negotiation)
The Conflict Minimisation Selection MechanismIssue analyser calculates the expected intersection size of this agents issues and opponents issuesUses attribute selection frequency information about the opponent
Selecting To Optimise Resource UseResource manager determines the expected deal price of an opponentChecks to see if the expected deal price is below reservationPrice profilesConcessionary flexibility
Combining the MechanismsSelection based on both conflict minimisation and resource optimisationMotivationally weighted by the worth of the goal and the worth of the resource
Preliminary EvaluationTested only on price minimisationCompare opponent selection against optimal selectionAgent learns to select the optimal
ConclusionsLittle autonomy apparent in pre-negotiation stageMotivation enables autonomous decision-making in dynamic negotiation settingsNew model of negotiation goals gives scope for motivation-based dynamic decision-makingCharacteristics of negotiation goal guides opponent selection.
End of Presentation
Dynamic domains and negotiation: changing needs, changing access to resources.
Many techniques and methods have been proposed to enable agent autonomy during the negotiation episode. Thus, the choice of strategy and tactics is often left to the agent at runtime given information about the opponent and the protocol in use and a set of preferences. Often the Justify motivation argument here!
When to move boxesUsing what methodFor how muchBrief intro to the warehouse domain
Quality = 3 different ways of moving the box. Each with a different risk of damaging the box.Skip?Skip this?What about negotiable versus fixed?