Next: Scenario Up: Tests with simulated Previous: Results

Adapting to Dynamic User Interests

To bolster the adaptive capabilities of the system, the system has been modeled as a team of specialized profiles. While each of the profiles is highly specialized in a narrow domain, the team as a whole reacts to changing user interests by controlling the numbers of each kind of specialized profiles. When user interests change, some of the profiles specialized to the old user interests are replaced by other profiles which serve the new user interest. This provides an optimal combination of specialization and adaptability.

In this section, the performance of the GA in evolving the population of specialized profiles is tested and the results presented.



MIT Media Lab - Autonomous Agents Group - agentmaster@media.mit.edu