Groupware that learns
Yasushi Nakauchi, T. Okada, Y. Anzai · 2002
The authors developed a groupware toolkit, called Michele, which supports asynchronous group work in distributed open environments. It is based on a multi-agent model of communication and allows a user to customize the interface by a specially designed language called MDL (multi-agent description language). The authors incorporate a learning mechanism in Michele. With the mechanism, Michele learns interactions between the user and the system, and stores learned knowledge in each user's environment. A document generation macro function has been added to MDL. With this macro function, cooperative work can be viewed as a set of document-driven procedures. Michele makes a decision tree for each query to the user when making new documents within the document generation macro function. The authors describe Michele and its agent description language MDL. They describe how to introduce a learning mechanism within Michele and explain the details of its implementation. An evaluation of the learning mechanism with a real example is described.>