Modelos de Contexto em PDAs
Authors
Abstract
Proactivity is one of the core subjects in Ubiquitous Computing. Inorder for a system to be proactive, it is necessary predict and anticipate the user's needs, in order to minimize distractions on a user's attention.
The task of anticipating the user's needs assumes that the system is
capable to accurately determine the user's context and act accordingly
in a precise and fast way.
Several works have already been proposed with the objective of
defining the context model as well the information that the model
should contain. But the same effort has not yet been applied to
context reasoning.
Considering this, the objective of the present work is to study which
are the learning algorithms more suitable for the proactivity
problem on Personal Digital Assistants (PDAs). For this purpose a
Sampling Application that acquires context information, based on the
user's location and on the presence of WiFi technology, was
developed with the objective of associating that information to the
applicational activity done on PDAs.
The results of the experience allow us to say that the clustering
algorithms, especially SimpleKMeans, are the most adequate for context
learning in proactivity problems on PDAs.