Amalthaea: An Evolving Multi-Agent Information Filtering and Discovery System for the WWW

  • Authors:
  • Alexandros Moukas;Pattie Maes

  • Affiliations:
  • Software Agents Group, MIT Media Laboratory, E15-305C, 20 Ames Street, Cambridge, MA 02139;Software Agents Group, MIT Media Laboratory, E15-305C, 20 Ames Street, Cambridge, MA 02139

  • Venue:
  • Autonomous Agents and Multi-Agent Systems
  • Year:
  • 1998

Quantified Score

Hi-index 0.00

Visualization

Abstract

Amalthaea is an evolving, multi-agent ecosystem for personalizedfiltering, discovery, and monitoring of information sites. Amalthaea'sprimary application domain is the World Wide Web and its main purpose is toassist its users in finding interesting information. Two differentcategories of agents are introduced in the system: filtering agents thatmodel and monitor the interests of the user and discovery agents that modelthe information sources.A market-like ecosystem where the agents evolve,compete, and collaborate is presented: agents that are useful to the user orother agents reproduce, while low-performing agents are destroyed. Resultsfrom various experiments with different system configurations and varyingratios of user interests versus agents in the system are presented. Finallyissues like fine-tuning the initial parameters of the system andestablishing and maintaining equilibria in the ecosystem are discussed.