Adapting Multimodal Interactive Systems to User Behaviour

completed

The subject of this thesis will be to develop techniques for multi-modal dialogue systems that are more user-friendly by making them adaptive to the needs and customs of the user. Many of today's interfaces, especially in the automotive environment, are very complex to use, while they should require as little attention as possible to avoid distraction. Automotive environments are characterized by a heterogeneous user group, which can be differentiated by several characteristics.

Experience and expertise: On the one hand, there are users who are accustomed to the interface by daily use, but demand an interface that can be used quickly. First-time users, on the other hand, require guidance while they explore and get to know the interface.
Situation and workload: When all the driver's attention is required for the primary task (e.g. while driving at high speed or in high traffic), an interface should distract the user as little as possible.
Overall demand: Moreover, different people have different demands toward dialogue systems. While some users want to use every detail of the system, others are satisfied when they know how to use the basic functionalities.
Specific needs: Especially elderly people require systems that are easy to use. Offering this group of users a system that provides less functionality, but in a way that they can use all of it, can be a better solution than offering lots of functionality that is not used due to the complex interface.

The aim of this thesis is to develop techniques to adapt dialogue systems to user behaviour, departing from the 'one size fits all' approach used today. The objective is to determine if and how dialogue systems can be adaptive and which parts of them are apt for adaptation. Different kinds of adaptations are possible, but only those that make the dialogue system more user friendly should be pursued. The utility of the developed approaches will be illustrated and evaluated with real users interacting with the developed prototype solutions.

• Adaptive and adaptable dialogue systems
• Modelling of adaptive systems

Master- & Bachelor-Theses

Publications