Adaptive information agents in distributed textual environments

Hypertext cnvironmcnts such as the Web are rich with both word and link cues that can be exploited by autonomous ngents performing distributed tasks on behalf of the user. This paper characterizes such environments and identifies the fcaturcs that are most useful and readily available. We dcacribe the adaptive representation of an ecology of retrieval agents who attempt to capture important features of their surroundings, and base their behaviors upon them. We PERCUSS how such a representation allows the agents to interact with the environments where they are situated. Agents cnn internalize words that are locally correlated with fitness, based on user feedback. They are shown to outperform nonndaptivn search by an order of magnitude. Furthermore, cnch agent learns new strategies at local time and space scnlcs, while the population evolves at a global scale.