Context Aware Communication, Terminal and User: User Terminal Interaction

1
Context Aware Communication, Terminal and User
User Terminal Interaction
CACTUS UseT Team
UseT Problem Setting In the near future devices,
services and content (DSC) will be accessible
through a network. Computing power and network
connectivity will be embedded in the environment.
A large number of sensors will generate huge
amounts of sensor data that can be used to infer
user intent, either implicit (e.g. user location)
or explicit (e.g. keyboard input). This is the
ubiquitous computing paradigm. Using the acquired
data in applications to create a smart
environment is called ambient intelligence. A
lot of research is being done regarding the
technological aspects of these paradigms, but it
remains unclear how a user will interact with
such an environment.
Perceived Consistency The focus of this task is
to improve the user experience of context aware
mobile devices and services by optimizing the
profit and minimizing the cost of user interface
adaptation. By means of experimentation and
scenario based evaluation user support is being
developed. The support should increase
consistency, transparency and the user's mental
model of context aware adaptive interfaces. This
increase will result in better performance,
learning and fallback recovery. A demonstration
of the first experiment is given at the TNO
stand. User Agent Interaction By adding agents
to the computing environment a number of issues
can potentially be resolved, e.g. keeping a
stable network, continuous user representation
etcetera. Furthermore, it allows other problems
to surface faster, i.e. the user interaction
problems can be isolated to the user agent
interaction. More specific, we study how the
combination of user constraints (cognitive,
social) and technical constraints of an ambient
intelligent environment gives rise to user
interaction problems. Service Matching If the
network can reach all devices in the environment,
the user will be overwhelmed by the sheer number
of agents and services directly available. We
developed a user-context sensitive agent that can
help the user finding the agent that best fits
his needs. In our current implementation the user
poses his needs in the form of a query in natural
language (English). See our poster "Service
Matching in Agent Space" for details. Personaliza
tion Information overload is the elementary
driving force behind the research into
personalization. In this project, the role and
functioning of a personalization service in a
pervasive computing environment is being
investigated. In this research, firstly, user
intent (interest) is inferred by assimilating
data from disparate sources (sensors, historical
interactions and etc.) in the ubiquitous
computing environment. Secondly, based on the
inferred user intent, interesting or useful
objects are provided in a large space of possible
options.
Users Agents Devices Services Content
Figure 1 Overview of UseT Multidisciplinary
approach Within the UseT research project,
several lines of research are combined Human
Factors, Interaction Design, Machine Learning and
Multi Agent Systems. Figure 1 provides a general
overview. From the User, via the Agents, to the
Services the following research topics are
pursued Willingness to Share, Perceived
Consistency, User Agent Interaction, Service
Matching and Personaliziation. The following
paragraphs detail these subprojects. Willingness
to Share This research focuses on the human side
of peer-to-peer technology. Peer-to-peer
technology depends heavily on the users
willingness to share resources. Our goal is to
develop and design new human centered technology
to support and stimulate the user to share more
resources. See our poster Willingness to Share"
for details.
Info H. de Ridder, h.deridder_at_io.tudelft.nl http
//www.cactus.tudelft.nl/