A user profiling architecture for textual-based agents

Abstract

Several intelligent agents have been developed in the last decade to help users with the vast amount of information available in the World Wide Web (WWW). Despite the efficiency of these agents depend on the knowledge they have about users, which is contained into user profiles, there are diverse considerations about what a profile should contain and how to construct it. Due to this fact, developers have to face the problem of, not only specifying the user profile content each time, but also its acquisition and adaptation to change in user interests. In this work we present an architecture which prescribes these aspects of the user profiling task. The goal of this architecture is to guide developers in the construction of agents involved with textual-based tasks. The results obtained from its application in a search agent, called PersonalSearcher, are also reported in this work.