A Software Engineering Framework for Context-Aware Service-Based Processes in Pervasive Environments - Advanced Design Approaches to Emerging Software Systems - page 106

Figure 1. The proposed context working definition

abstraction or constructs that model the context
variability. Indeed, different context knowledge
could be extracted from the context repository by
focusing on different views of the context informa-
tion. For example, in the smart meeting room, a
seat may be equipped with light and temperature
sensors to reason about its occupation. The seat
could be either free or occupied. Two occupation
variants may be identified: occupied by object and
occupied by a person. These variants represent
two facets to the same fact. Another example
of context variability is the context information
classification. For instance, the room temperature
could be classified as low, moderate and high ac-
cording to some specified temperature ranges; but
these ranges could be different if the room type
is a sitting room or a sauna. To the author's best
knowledge, the existing approaches do not provide
application developers with software constructs
through which a view-based customization of the
context knowledge could be expressed.
As one of the successful research directions
in software engineering, software product line
research could contribute to the context modeling.

Commonality and variability management tech-
niques from software product line can be applied to
handle context variabilities for customization and
adaptation (See “Context as a Dynamic Product
Line”). Therefore, in this chapter we explore the
synergy between feature modeling and context
modeling. On the other hand, feature modeling is
a key concept in product line engineering. Thus,
the feature model of the system context will be
considered as a composition of segmented context
features models; each of which models a part of
the whole context. Based on the context feature
model, specific context −member of a product
line− can be constructed by composing features
from context information.
In this section we focus on dealing with con-
text variability from the application requirement
perspective. The proposed approach does not
model the context information itself by using
feature model as the feature models are less pow-
erful than ontologies, and are more appropriate
for expressing a subset of what ontologies can
express (Czarnecki et al. 2006). Instead, the aim
is to represent the context information from the