On Short-Term and Long-Term Memory for Brahms Agents

Maarten Sierhuis and Ron van Hoof

Brahms is a multi-agent modeling language for developing distributed multi-agent systems, as well as multi-agent simulations. One of the issues we face developing Brahms agents is that the belief-set of an agent increases over time, slowing down the agent’s processing capability. Another issue is how an agent can "remember" what happened in the past. How can an agent’s memory be modeled sufficiently so that it can remember what activities were performed and how to continue a task that was not finished the day before? How does the agent remember the context of activities from the past? A design and use of agent memory is described that includes both short-term and long-term memory based on context, as well as the use of such memory to handle the growing set of beliefs. The interaction between short- and long-term memories over time is also discussed.


This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.