Polymorphic Agents for Modeling E-Business Users

摘要

Considering the huge variability of users in e-Business as well as the increasing need for intelligent tools to meet their requirements, the paper presents polymorphic agents as a means to model, refine and update dynamically user profiles. In detail, user profiles are adapted continuously corresponding to different stages (of activity) and/or stances (of immediate interest). After an evaluation of the comprehensive related work in this area the paper presents an overall approach based on polymorphic agents, conceptually modeled as generic entities and instantiated for each specific purpose (related application or sub-goal of the same one); the main mechanism used is 'flexible cloning'. Based on this groundwork, the architecture of profiler agents as well as an experimental model are outlined. It is considered that this approach can be easily extended to any kind of agents involved in e-Business applications.