Spawning information agents on the web

Network bandwidth limitations must be considered by information agents when retrieving information from remote sources. There are cases where it should be more efficient (considering network constraints) that an agent will perform its activity on the remote server. This can be performed by spawning an agent on that server. Agent spawning is an appropriate means for resolving agent overload problems as well. Agents in a multi-agent system may face situations where tasks overload their computational capacities or do not fit their capabilities. Usually, this problem is solved by passing tasks to other agents or agent migration to remote hosts. Agent spawning is a more comprehensive approach to balancing local agent overloads. According to our paradigm, agents may spawn, pass tasks to others, die or merge. We discuss the mechanisms required for deciding upon and performing agent spawning. The requirements for implementing a spawning mechanism and the advantages of using it are addressed.