1 / 31

Rustam Vahidov John Molson School of Business Concordia University Montreal, Quebec, Canada

Adopting Agent-Based Situated Decision Support Framework for Managing One-to-Many Negotiations with Multiple Potential Agreements. Rustam Vahidov John Molson School of Business Concordia University Montreal, Quebec, Canada. Introduction. e-Commerce and e-Negotiations Negotiation Support

elton
Download Presentation

Rustam Vahidov John Molson School of Business Concordia University Montreal, Quebec, Canada

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. Adopting Agent-Based Situated Decision Support Framework for Managing One-to-Many Negotiations with Multiple Potential Agreements Rustam Vahidov John Molson School of Business Concordia University Montreal, Quebec, Canada

  2. Introduction • e-Commerce and e-Negotiations • Negotiation Support • Software agents • Automated negotiations • Agent-assisted negotiations

  3. E-Negotiation Technologies • Communication • Electronic message exchange • Analytical • Negotiation support systems • Analytical toolbox • Agents • Agent-integrated negotiation support • Assists (guides) negotiators in the process

  4. Automated negotations • One-to-one, e.g. • AuvtionBot, Kasbah & Tête-à-Tête • (Faratin, Sierra, Jennings, 2002) • “Smart” similarity-based negotiation strategy • (Sycara, 2006) • Complex preferences, uncertainty with regard to opponent’s and own preferences • Mapping business policies and contexts to negotiation goals, strategies, plans, and decision-action rules (Li, Su & Lam, 2006)

  5. Automated negotiations • Opponent profiling • Modeling opponent attitude (Lee, 2004) • Bayesian belief revision (Zeng & Sycara 1998) • Probabilistic influence diagrams (Mudgal & Vassileva 2000) • Predicting opponent moves with ANN (Carbonneau et al, 2006)

  6. Agent-assisted negotiations • Preference elicitation • Information search & retrieval • Offer generation • Offer critique • Counter-offer evaluation & critique • Opponent modeling • Aspire (Kersten & Lo, 2001) & eAgora (Chen, Kersten & Vahidov, 2005) systems

  7. One-to-Many (multi-bilateral) Negotiations • Analysis of alternatives through fuzzy set-theoretic model (Van de Walle, Heitsch & Faratin, 2001) • Single coordinating agent, multiple negotiating agents (Rahwan, Kowalczyk & Pham, 2002) • Coordinator, multiple agents communicating intermediate deals, leveled decommitments (Nguyen & Jennings, 2004) • Game & decision-theoretic approach to support both sides (Lu, feng & Jiang, 2005)

  8. Multi-bilateral negotiations with multiple possible agreements • One party negotiating with N other parties for M possible agreements • AutONA: one party negotiates with multiple suppliers to find distribution of quantities supplied (Byde & Chen, 2003) • Load balancing negotiations with multiple power consumption agents (Brazier et al. 2000)

  9. Objective • Propose framework for managing one-to-many operational-level negotiations with multiple possible agreements • Human decision maker in control of the overall process • Managing negotiating businesses • Relating low-level operational negotiations to business goals & objectives

  10. Situated Decision Support (Vahidov & Kersten, 2004)

  11. Analogy with AI vs. IA

  12. Situated Decision Support (“Decision Station”) • DSS kernel • (traditional toolbox) • Active user interface • (supporting problem-solving by the user) • Sensors • (non-trivial information & alert delivery) • Effectors • (non-trivial decision implementation) • Manager • (limited autonomy) • Implementation: personal finance management system

  13. Adopting SDSS for managing multiple negotiation processes • Effectors: • Automated or agent-supported conduct of negotiations • Opponent profiling • Sensors • Delivery of relevant information, e.g. market indicators & news filtering • Manager • Monitoring overall performance & making adjustments to preferences, strategies, reservation values on the basis of agent’s performance and market information within limits specified by the user. • Generating recommendations for the user

  14. Adopting SDSS for managing multiple negotiation processes • Models • Predictive & simulation models for estimating number of final agreements, profits, resource consumption, opportunities lost, etc. • Optimization models • User • Makes judgment in setting current limits for intervals on decision variables, based on model outputs, manager recommendations, business goals & policies, and external information

  15. Negotiation station Judgmental/ Strategic Planning/ Tactical Reactive/ Operational

  16. Components • Effectors • Adapt to opponent profile • Follow provided preference structure and reservation levels • Manager • Monitors performance of effectors and compares the outcomes with goals and resources for a given period • Makes adjustments to reservation levels and issue preferences subject to constraints • Sends alerts and makes recommendations to decision maker if goals deemed unachievable

  17. Components • Decision maker • Utilizes models to set goals and limits for the autonomous negotiations throughout the process • Exercises judgment based on knowledge of the market, possible external effects, company policies, and risk attitude

  18. Feasibility • Maybe making tradeoffs or concessions • Possible example cases: used auto sales, travel packages (e.g. Priceline, Hotwire), selling through exchanges, auctions. • Concession: may be given based on time spent in negotiations

  19. Case • Condo rental case tested with eAgora (Vahidov, Chen & Zhen, 2005) • Issues: Price, parking, cleaning, deposit, duration • Multiple units • Simulations

  20. Simulations • Very preliminary results • Assumptions: • 20 days time horizon • 80% customers price-sensitive • Each negotiation finished in a single day • Agreements are Nash solutions (e.g. agents follow “smart” strategy by Faratin et al.)

  21. Effect of price reservation level adjustments by manager on agreements

  22. Effect of parking importance adjustment on parking spaces rented

  23. Comparison of profits made by fixed pricing vs. negotiations without & with manager’s adjustments

  24. Comparisons under steadily increasing market price

  25. Price adjustments under increasing market price

  26. Sudden Price increase in period 10

  27. Decision maker’s intervention in period 10

  28. Simulations • Under increasing price scenario market price increased from $650 to $745 • In all cases agent-based systems outperformed fixed pricing policy • In case of increasing prices agent systems did even better

  29. Summary & Conclusions • An agent-based systems for managing negotiations • Combining automated action with human judgment • Preliminary simulation results

  30. Future work • Prototype implementation • Realistic assumptions • Human subject experiments • Support of vague decision making?

More Related