What is Knowledge Based Systems

Today’s e-commerce environment requires that interactive systems exhibit abilities such as autonomy, adaptive and collaborative behavior, and inferential capability. Such abilities are based on the knowledge about users and their tasks to be performed (Raisinghani, Klassen and Schkade, 2001). To adapt users’ input and tasks an interactive system must be able to establish a set of assumptions about users’ profiles and task characteristics, which is often referred as user models. However, to develop a user model an interactive system needs to analyze users’ input and recognize the tasks and the ultimate goals users trying to achieve, which may involve a great deal of uncertainties. Uncertainty refers to a set of values about a piece of assumption that cannot be determined during a dialog session. In fact, the problem of uncertainty in reasoning processes is a complex and difficult one. Information available for user model construction and reasoning is often uncertain, incomplete, and even vague. The propagation of such data through an inference model is also difficult to predict and control. Therefore, the capacity of dealing with uncertainty is crucial to the success of any knowledge management system. Currently, a vigorous debate is in progress concerning how best to represent and process uncertainties in knowledge based systems. This debate carries great importance because it is not only related to the construction of knowledge based system but also focuses on human thinking in which most decisions are made under conditions of uncertainty. This chapter presents and discusses uncertainties in the context of user modeling in interactive systems. Some elementary distinctions between different kinds of uncertainties are introduced. The purpose is to provide an analytical overview and perspective concerning how and where uncertainties arise and the major methods that have been proposed to cope with them.