Queries that are not indicative of real information needs are amajor problem for information retrieval systems. In this work we studyhow individual learning helps adaptive agents, when searching forinformation in a distributed environment, to modify incomplete queriesin order to improve their retrieving performance. Two learningprocedures, occurring in two different levels, are proposed and theireffect is studied in several situations. Preliminary results show thatchanges induced by learning in the query vector of adaptive agents,provide an important advantage and enable them to make correct decisionsabout how to deal with this problem
展开▼