Agent Academy

A Data Mining Framework for Training Intelligent Agents

 agents

Acronym AA
Title AgentAcademy: A Data Mining Framework for Training Intelligent Agents
Funding Source European Commission – IST program
Dates 11/01-4/04
Amount €3,100,467 (our part is ~€880,000)
Role Project Coordinator

Description

The Agent Academy (AA) framework operates as a multi agent system, which can train new agents or retrain its own agents in a recursive mode. A user (a physical or virtual entity) issues a request for a new agent as a set of functional specifications. The request is handled by the Agent Factory, a module responsible for selecting the most appropriate agent type and supplying the base code for it. This untrained agent (UA) comprises a minimal degree of intelligence, defined by the software designer. It enters the Agent Training (AT) module, where its world perception increases substantially during a virtual interactive session with an agent master (AM). Based on the encapsulated knowledge, acquired in the knowledge extraction phase, an AM can take part in long agent-to-agent transactions with the UA. This process may include modifications in the agent’s decision path traversal and application of augmented adaptivity in real transaction environments. The core of the agent academy is the Agent Use Repository (AUR), a collection of statistical data on prior agent behavior and experience. Building AUR will be a continuous process performed by a large number of mobile agents and controlled by the data acquisition module. It is on the contents of AUR that the Data Miner, the data mining implementation module, performs agent type classification and association rules extraction for the decision making process, in order to augment the intelligence of the AM in the training module. A large part of an agent’s intelligence handles the knowledge acquired by the agent since the beginning of its social life through the interaction with the environment it acts upon. After the training is complete, the now intelligent agent, armed with tools for reporting its behavior to the AA, is released to the world. The AA is contiuously integrated, as it receives feedback from mobile agents roaming the web, updating its agent use repository and refining its data mining and AT techniques.

Project Milestones

The project’s milestones and results are:

  • the integration of a generic intelligence embedding training framework
  • the establishment of novel methodologies that enhance Intelligent Agents (IAs) intelligence
  • the creation of a large repository of data on IA use and behavior
  • the statistical analysis of resulted data and extraction of knowledge about IA’s intelligence and learning process
  • adoption of IAs added value services by enterprises and research organizations

Project Objectives

The overall objective of the proposed project is the development of an integrated framework that enables the improvement of Intelligent Agents (IAs) ability of learning, making decisions and create inferences, based on Data Mining (DM) techniques. The successful outcome of this effort is expected to propagate the use of IA related technologies into both business practices and personal use. Specific objectives are:

  • the development of novel methodologies to enhance IAs intelligence
  • the development of tools for assembling and maintaining a large repository of data on IA use and behavior, as well as DM techniques for knowledge extraction about IAs behavioral characteristics
  • the improvement of the quality of services provided by enterprises solutions based on IAs applications
  • the exploitation and early adoption of project results across Europe and
  • the dissemination of the project results to make them visible to European and international research and industrial communities

Project Website