Even with assistive communication technology, interactive conversation is extremely difficult for users with severely limited mobility and loss of speech. Input to such devices is painfully slow and subject to high error rates with the resulting output not reliably reflecting the true intentions of the user. Conversational prediction has been incorporated into assistive systems to help speed up communication but could be further improved by considering the contextual interaction between the user and conversant. Contextual information applied to user profiles can greatly enhance conversational prediction and increase a severely disabled user's control over his or her complex world. We present a framework that integrates a rich profile of the user, a model of the user's environment, and actors on that environment. To test the validity of the framework, we develop a set of profiles and apply them in two different scenarios. Initial results show that the context-aware user profiles can increase both the accuracy and speed of the communication.
[1]
Robert T. Craig,et al.
Conversational coherence : form, structure, and strategy
,
1983
.
[2]
Ann Copestake.
Applying Natural Language Processing Techniques to Speech Prostheses
,
1996
.
[3]
D. Sperber,et al.
Relevance: Communication and cognition, 2nd ed.
,
1995
.
[4]
Alan F. Newell,et al.
Prediction and conversational momentum in an augmentative communication system
,
1992,
CACM.
[5]
Susan T. Dumais,et al.
Personalized information delivery: an analysis of information filtering methods
,
1992,
CACM.
[6]
Anthony Jameson,et al.
Modelling both the Context and the User
,
2001,
Personal and Ubiquitous Computing.
[7]
Thomas W. Malone,et al.
Intelligent Information Sharing Systems
,
1986
.
[8]
Peretz Shoval,et al.
Information Filtering: Overview of Issues, Research and Systems
,
2001,
User Modeling and User-Adapted Interaction.