Information state and dialogue management in the TRINDI dialogue move engine toolkit

We i n troduce an architecture and toolkit for building dialogue managers currently being developed in the TRINDI project, based on the notions of information state and dialogue move engine. The aim is to provide a framework for experimenting with implementations of diierent theories of information state, information state update and dialogue control. A n umber of dialogue managers are currently being built using the toolkit, and we present a detailed look at one of them. We believe that this framework will make implementation of dialogue processing theories easier, also facilitating comparison of diierent t ypes of dialogue systems, thus helping to achieve a prerequisite for arriving at a best practice for the development of the dialogue management component o f a s p o k en dialogue system. 1 I n troduction We discuss a candidate model for best practice in the development of the dialogue management component of a spoken dialogue system. At this early stage in development o f s u c h systems, there is a wide range of current practice, depending in part on the diierent tasks that such systems are put to, but also on external factors, such as preference and background of the developers. In order to determine what the best practice is, if indeed there can be a single model that is useful for the whole range of such systems, it is necessary to be able to compare and evaluate these components. A big diiculty is that not only are the dialogue management c o m p o n e n ts themselves radically diierent, but the systems of which they are a part are also quite heterogeneous, making principled comparison of the components diicult, if not impossible. What we propose is a view of dialogue management functions in terms of information state. Key to this approach i s i d e n tifying the relevant aspects of information in dialogue, how they are updated, and how updating processes are controlled. This simple view can be used to compare a range of approaches and speciic theories of dialogue management within the same framework (as well as facilitating hybrid approaches). Use of the framewo r k w i l l t h us allow comparison to determine empirically which is the best practice. The term information state of a dialogue represents the information necessary to distinguish it from other dialogues, representing the cumulative …