Multimodal content adaptation for heterogeneous devices
Songsak Channarukul, Susan McRoy, Syed Sumera Ali · 2005
This thesis considers an approach to multimodal content adaptation that integrates constraints from three separate sources: (1) the user, (2) the device, and (3) the domain. This integration of constraints sets this thesis apart from existing research. In the past, user and device constraints have rarely been taken into account together in any significant way. In this thesis, I present a new computational framework for finding the optimal combination of content, modality, and layout for individual users, their current device and computing environment, and the domain being presented. Thus framework has been implemented as Multiface, a real-time, interactive agent. Its plans have been developed in the JAM language and uses Huber's [28] generic planning agent. The core of my approach is a completely plan-based description of how to adapt content and modalities from an annotated document and models of the user and device. This approach makes use of domain specific and domain independent information. Domain specific rules are created from an annotated document that defines content units, structure, and semantic relations among those units. Domain independent rules represent general knowledge about user models, device characteristics, modality relations, and high level pedagogy. I will utilize them to generate adaptive multimodal contents for heterogeneous devices.