The COSIMA prototype for multi-objective bargaining

  • Authors:

  • Stefan Fischer;Werner Kießling;Stefan Holland;Michael Fleder

  • Affiliations:

  • University of Augsburg, Augsburg, Germany;University of Augsburg, Augsburg, Germany;University of Augsburg, Augsburg, Germany;University of Augsburg, Augsburg, Germany

  • Venue:
  • Proceedings of the first international joint conference on Autonomous agents and multiagent systems: part 3
  • Year:
  • 2002

Quantified Score

Hi-index 0.00

Visualization

Abstract

Multi-objective bargaining is a challenging business model for B2C or B2B Internet applications. Our visual bargaining agent COSIMA builds on cooperative database technology like Preference SQL or Preference XPATH, which can efficiently implement \linebreak[4] multi-dimensional search engines with a Pareto-optimality semantics to find best alternative offers. COSIMA is a smart avatar with dynamically generated speech output. We have implemented several well-known bargaining strategies for a sample e-shop, where the customer can bargain on more attributes than just the price. The life-like appearance of COSIMA is supported by randomizing her bargaining tactics and by expressing visual emotions during bargaining. She can also learn and adapt dynamically to various bargaining preferences of the customers. The COSIMA prototype is carefully engineered by means of component-ware and IT-standards including SQL, XML and Java. This makes it scalable even to very large e-shop databases. Measurements show that this advanced user interface of COSIMA requires only a small overhead of a few seconds, including the bidding process. Thus real-time performance using state-of-the art software components and hardware is feasible. A preliminary user feedback has been very encouraging. COSIMA gave her debut to a large professional audience at the computer fair CeBIT 2002.