1 / 22

Automated negotiations

Automated negotiations. The best terms for all concerned Tuomas Sandholm sandhom@cs.wustl.edu. Overview. Automated negotiations Promising application areas Reallocation among agents Optimization Agent-to-agent vs. centrally mediated eMediator: features. Automated negotiations.

tavita
Download Presentation

Automated negotiations

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Automated negotiations • The best terms for all concerned • Tuomas Sandholm • sandhom@cs.wustl.edu

  2. Overview • Automated negotiations • Promising application areas • Reallocation among agents • Optimization • Agent-to-agent vs. centrally mediated • eMediator: features

  3. Automated negotiations • Negotiation: key component in e-commerce • Auto: computational agents • Represent real-world parties • Find & prepare contracts • Benefits …

  4. Auto negotiation benefits • Save human negotiation time • Better at finding deals in combinatorially & strategically complex situations • Not at cost to other parties • Rapidly find solutions that optimize for all

  5. Promising application areas • Retail e-commerce • Electricity markets • Bandwidth allocation • Manufacturing planning & scheduling in subcontracting networks • Distributed vehicle routing among independent dispatch centers • Electronic trading of financial instruments

  6. Reallocation among agents • Task reallocation among agents: key type of negotiation • Allocate tasks to agents that handle them least expensively • Marginal cost-based method for automated task reallocation • Reallocate all types of items ….

  7. Items to reallocate • Tasks • Financial instruments • Hours of electricity

  8. Cost-based reallocation • Agent takes task from another as long as it is paid more by other agent than what costs to handle • Agent gives task to another agent as long as it doesn’t have to pay more than it would cost to handle • TRACONET: TRAnsportation Cooperation NET

  9. TRACONET • Automated delivery of task reallocation among freight companies • 1st distributed automated negotiation among self-interested agents • Each agent • Own Unix process • Represents single company

  10. TRACONET (cont.) • Agent takes on delivery tasks • Gives out tasks • Recontract-out tasks previously contracted-in

  11. Optimization • Negotiation can get stuck in local optimization • Task allocation suboptimal • No original contract (O) profitable • To solve: new contract types …

  12. New contract types • C: cluster • Exchange multiple tasks for payment • S: swap task for another + sidepayment • M: multiagent • More than 2 parties in same contract • OCSM …

  13. OCSM … • Combine previous into atomic contract • Guarantee globally optimal allocation • Through finite number of contracts

  14. Agent-to-agent vs. centrally mediated • Original: agent-to-agent • Auction server ==> centrally mediated variant • Agents send bids and tasks • ==> Combinatorial auctions • Allow users to express interrelated valuations of items

  15. eMediator: features • E-commerce server • Also services other than auctions • Combinatorial bidding • Bidding via price-quantity graphs • Mobile agents • Determine winners of combinatorial auction • Identify profitable contacts for all

  16. Optimal winner determination • Computationally complex • ==> Added highly optimized search-based matching algorithm to solve problem

  17. Price-quantity graphs • User can express continuous preferences

  18. Mobile agents • User have agents participate in auctions while users disconnected from Internet • Mobile agents execute on agent dock on or near auction host • ==> Reduce network latency • Key issue in time-critical bidding • Mitsubishi’s Concordia agent dock

  19. Concordia • Give mobile agents safe execution platform • Bid • Set up auctions • Travel to other auction sites • Observe activity at various auctions • More later

  20. HTML interface • Users instruct agents • Automatic generation of Java code for mobile agents before launching

  21. Contract management • Usually: binding • ==> Can’t undo old commitments to accommodate new events • E.g., tasks more costly than anticipated • New offers more lucrative • Alleviate: Leveled commitment contracting protocol …

  22. Leveled commitment contracting protocol • Agents accommodate future events • Option of unilateral decommit • Decommitment penalty

More Related