Implicit Training of Virtual Shopping Assistants in 3d Electronic Institutions

The growing demand for shopping assistants in E-Commerce was identified by many researchers. Some retailers try to address this need by introducing totally autonomous agents; others make extensive use of human resources, shifting operators from telephone lines to chat-based interactions with online customers. The recently developed 3D Electronic Institutions methodology provides facilities to conveniently combine these two approaches. Initially, an autonomous agent tries to deal with customer’s requests. When the limitations of its intelligence are reached a human operator takes over and satisfies the "out-of-scope" inquiry. At the same time, the agent observes the human operator and learns how to handle similar inquires in the future. We argue that this learning aspect can be realized by means of 3D Electronic Institutions and believe that this "agent training" will be feasible and far more successful than it is possible in nowadays form-based E-Commerce solutions.