=Paper=
{{Paper
|id=Vol-210/paper-16
|storemode=property
|title=Quality Extensions and Uncertainty Handling for Context Ontologies
|pdfUrl=https://ceur-ws.org/Vol-210/paper16.pdf
|volume=Vol-210
|dblpUrl=https://dblp.org/rec/conf/ecai/PreuveneersB06
}}
==Quality Extensions and Uncertainty Handling for Context Ontologies==
https://ceur-ws.org/Vol-210/paper16.pdf
Quality Extensions and Uncertainty Handling
for Context Ontologies
Davy Preuveneers and Yolande Berbers 1
Abstract. Context, by nature, involves real world entities and is 2 RELATED WORK
therefore subject to uncertainty and inaccuracies. Ontologies are of-
ten used to model context in a formal way in order to achieve a In this section we focus on those contributions on quality of con-
shared semantic understanding of concepts and the relationships that text and uncertainty management for mediation of ambiguous con-
hold among them. However, they lack support for representing am- text that are most related to the work presented in this paper. This
biguous context and appropriate comparison algorithms. As such, work is based on the ideas presented in Buchholz et al. [11], where
context-aware applications may make the assumption that the con- the authors identify parameters that quantify the Quality of Context
text they use is completely accurate. In this paper we propose a sim- (QoC) and the inevitable uncertainty of sensed values for individual
ple and lightweight yet generic approach to extend context ontolo- context properties:
gies with quality of context properties and discuss the use of these • Precision: describes how sharply defined a measurement is stated
quality properties for context ontology matching under uncertainty and what the difference is with the actual value in the real world.
using fuzzy set theory. We illustrate the proposed extensions and un- • Probability of correctness: estimates how often the context infor-
certainty mechanisms with a small example where uncertain spatio- mation is unintentionally wrong due to internal errors.
temporal coverage is combined with other contextual properties. • Trust-worthiness: describes the reliability of the entity that may
have persistently provided incorrect information in the past.
• Resolution: describes the granularity of the information and the
1 INTRODUCTION inability to offer information with a finer detail.
Context-awareness has been drawing much attention from re- • Up-to-dateness: describes the age of information which can be
searchers in the ubiquitous and pervasive computing domain [12] as used to decide on the temporal relevancy in a particular situation.
context has become a key ingredient to create a whole range of smart Henricksen et al. [6] explore the problem of imperfect context in-
entertainment and business applications that are more supportive to formation and characterize the following four types and sources of
the user. Context [4] has been defined as any information that can imperfect context information: Unknown, Ambiguous, Imprecise and
be used to characterize the situation of an entity. Humans take this Erroneous. The first two types of imperfection are new, whereas the
context information into account rather intuitively, whereas context- latter two types combine several Quality of Context properties on the
aware applications require an explicit model to take advantage of list of the work by Buchholz et al..
context information for non-intrusive decision making and adapta- In [7] Parsons describes qualitative methods for reasoning with
tion [9]. Imperfections in the context data can cause incorrect or unin- various types of imperfect information and argues that qualitative
tended application behavior as relationships between similar context methods have the advantage to not require precise numerical infor-
properties become uncertain. For example, the precision of a coordi- mation, but instead to rely on abstractions such as interval values and
nate based positioning system is required to decide whether a given information about how values change.
position matches with a location such as ‘at the office’. Chalmers et al. show in [1] how context can be formulated in the
In this paper we propose to extend context ontologies with quality presence of uncertainty using interval arithmetic for numerical con-
of context properties and discuss a lightweight and generic approach text values, and analogously using trees with abstract values for con-
for matching under uncertainty that is simple enough to be imple- text ontologies. The authors define within and overlap relationships
mented and used on resource constrained devices, such as PDAs. between actors and context objects both for numerical and abstract
The remainder of this paper is organized as follows. In section 2 values in order to compare context information.
we describe related work on quality of context and reasoning with
uncertainty. Section 3 discusses how quality of context aspects are
introduced into our context ontology. Section 4 describes the use 3 EXTENDING ONTOLOGIES WITH
of membership functions based on the concept of fuzzy set theory QUALITY PARAMETERS
to achieve advanced matching mechanisms for context ontologies in Ontologies and the Web Ontology Language (OWL) are very pop-
the presence of uncertainty. In section 5 we conduct an experiment ular for a systematic arrangement of context concepts and the rela-
illustrating uncertain spatio-temporal coverage combined with other tionships that hold among them [10, 2, 5]. In our previous work [3]
contextual properties to validate the matching mechanisms in more we defined an OWL context model specifying User, Platform, Ser-
advanced scenarios. We conclude with section 6. vice, Environment and related concepts to provide a shared semantic
1 Katholieke Universiteit Leuven, Belgium, email: {davy.preuveneers, understanding for context-driven adaptation of mobile services. Our
yolande.berbers}@cs.kuleuven.be context system [8] is able to gather and interpret this information. In
rdfs:domain rdfs:range
fV(x)
Membership
rdf:Property
c*t r
r
owl:ObjectProperty owl:DatatypeProperty
precision trust
correctness resolution
0 p*(v-r) v-r v v+r 2v-p*(v-r) X
QualityExtension ...
Figure 2. Membership function for a single sensed value with given
Quality of Context parameters
owl:QXObjectProperty owl:QXDatatypeProperty
fC(x)
Membership
class property instance
Figure 1. Extending the OWL language with QoC properties
the case of uncertainty in the gathered information, the context-aware 0 X
system needs context quality parameters in OWL in order to deter-
mine a high confidence of correctness of matching context informa-
Figure 3. A fuzzy set C as a averaged sum of single fuzzy sets
tion. We will now show how the Quality of Context (QoC) param-
eters discussed in [11] are modeled by means of two new property
types, QXObjectProperty and QXDatatypeProperty. Both property need appropriate matching algorithms that take into account the im-
types inherit from the DatatypeProperty and ObjectProperty OWL perfect nature of context when taking appropriate actions. In this
language constructs, as well as from a self-defined class QualityEx- section, we will show how we use concepts of fuzzy set theory of
tension which models the Quality of Context parameters precision, Zadeh [13] and define membership functions based on the quality of
correctness, trust and resolution as DatatypeProperties: context parameters defined in the previous section.
4.1 Modeling a fuzzy context concept
In classical set theory the membership of an element to a set can be
clearly described. In fuzzy set theory, an element belongs to a set with
a certain possibility of membership. Age is a typical example of a
fuzzy concept. There is no single quantitative value or clear boundary
... defined for the term young: age 25 can be young for some, while age
30 can be young for others. However, age 1 is definitely young, while
age 100 is is definitely not young.
We can model the membership function for a single sensed value
using the Quality of Context parameters in a similar way. Assume
a sensed value v has a precision p, a probability of correctness c,
a trust-worthiness t and a resolution r, with 0 ≤ p, c, t ≤ 1, then
we define the following symmetric membership function fV (x) with
See Figure 1 for an overview of the property inheritance hierarchy. x ∈ X for the sensed value v as in Figure 2. Note how the Quality
The QoC parameters of e.g. a sensor that instantiates the temperature of Context parameters change the crisp sensed value into an interval
concept in our context ontology [3] are modeled as follows: with a particular symmetric shape of the fuzzy set.
4.2 Aggregation and matching of fuzzy concepts
If a contextual concept C is defined by set of N measured values
95 vi then we can improve the accuracy of its membership function by
100
1 using the aggregated membership of this concept fC (x) with x ∈ X
... defined as the averaged sum of fVi (x):
P
fVi (x)
fC (x) = with x∈X
4 MATCHING IN THE PRESENCE OF N
UNCERTAINTY WITH FUZZY SETS
For example, our WiFi location sensor uses multiple Received Sig-
In the real world context information can be vague, imprecise, un- nal Strength Indication (RSSI) values as a distance measurement to
certain, ambiguous, inexact, or probabilistic in nature. We therefore known access points and models them as fuzzy sets. An example of
such an averaged sum of these fuzzy sets is shown in Figure 3. Note 6 CONCLUSION
that the aggregated fuzzy set is no longer symmetric.
We define a match between two sensed values with fuzzy sets A In this paper we have proposed a simple and lightweight extension
and B and membership functions fA (x) and fB (x) based on the to the OWL language to model quality of context properties in order
to deal with ambiguous and imperfect context information. We have
T of fuzzy sets A and B. The intersection [13] is aV
intersection fuzzy set
discussed the use of these quality parameters in automated uncer-
C = A B with a membership function fC (x) = fA (x) fB (x)
which is defined as follows: tainty reasoning to achieve more advanced matching mechanisms for
V context ontologies. This automated uncertainty reasoning was based
fC (x) = fA (x) fB (x) = Min[fA (x), fB (x)] on concepts of fuzzy set theory. We have illustrated the proposed
Two fuzzy concepts match if their overlapping area is larger than a ontology extensions and the fuzzy comparing algorithms with small
user-defined and context-specific threshold α: examples which included spatio-temporal coverage as fuzzy sets.
R The proposed matching mechanisms are still a work in progress,
X
fC (x) but worked as expected for the examples. Difficulties are assumed to
0≤α≤ R R ≤1 with x∈X
M in[ X fA (x), X fB (x)] arise when the number of fuzzy sets involved in a single contextual
condition is going to increase. We therefore will further continue to
Of course, when one of the membership functions is f (x) = 0 or
refine the membership functions by including the likelihood of con-
when the overlap is zero, then there is no need to calculate this ratio.
text information in order to reduce to possible scenarios that may
match under particular circumstances. One improvement that may
5 EVALUATION proof to be useful is the inclusion of likelihood of events. This will
This subsection discusses the scenario used for a preliminary evalu- better differentiate the likelihood of fuzzy matches.
ation of the uncertainty mechanisms for matching context informa-
tion. A PDA enabled with WiFi networking is used for Received Sig- REFERENCES
nal Strength Indication (RSSI) based location-awareness. The com- [1] D. Chalmers, N. Dulay, and M. Sloman, ‘Towards reasoning about con-
puter science building has about 100 offices, labs and meeting rooms text in the presence of uncertainty’, in Proceedings of Workshop on Ad-
and is equipped with 7 access points for wireless Internet access on vanced Context Modelling, Reasoning And Management at UbiComp
all 5 floors. In the first step we trained the system by walking around 2004, (2004).
[2] Harry Chen, Tim Finin, and Anupam Joshi, ‘An ontology for context-
in the building and taking about 10 measurements for several offices. aware pervasive computing environments’, Special Issue on Ontologies
We determined the Quality of Context parameters based on a long for Distributed Systems, Knowledge Engineering Review, (2003).
test run while remaining at the same location. We looked for outliers [3] D. Preuveneers, et al., ‘Towards an extensible context ontology for am-
in the sampled data, calculated the mean and variation in the data and bient intelligence’, in Second European Symposium on Ambient Intelli-
gence, volume 3295 of LNCS, pp. 148 – 159, Eindhoven, The Nether-
estimated the values of the QoC parameters as follows:
lands, (Nov 8 – 11 2004). Springer.
• Precision: 95% [4] Anind K. Dey, ‘Understanding and using context’, Personal Ubiquitous
• Probability of correctness: 90% Comput., 5(1), 4–7, (2001).
• Trust-worthiness: 100% [5] T. Gu, X. H. Wang, H. K. Pung, and D. Q. Zhang. An ontology-based
context model in intelligent environments. In Proceedings of Commu-
• Resolution: 3 dBm nication Networks and Distributed Systems Modeling and Simulation
Using this information, the average fuzzy set for each of the ac- Conference, San Diego, California, USA, January 2004.
cess points that were seen in a particular office was calculated. Af- [6] Karen Henricksen and Jadwiga Indulska, ‘Modelling and using imper-
fect context information’, in PERCOMW ’04: Proceedings of the Sec-
ter ordering the overlap ratios by decreasing order, and selecting the
ond IEEE Annual Conference on Pervasive Computing and Communi-
fuzzy set with the highest overlapping ratio, the locations matched, cations Workshops, p. 33, Washington, DC, USA, (2004). IEEE Com-
although non of the new RSSI measurements was exactly equal to a puter Society.
previously encountered measurement at the same location. [7] Simon Parsons, Qualitative Methods for Reasoning under Uncertainty,
In a second test scenario which illustrates spatio-temporal cover- The MIT Press, August 2001.
[8] Davy Preuveneers and Yolande Berbers, ‘Adaptive context manage-
age, my PDA informs the instant messaging client on my desktop ment using a component-based approach’, in Proceedings of 5th IFIP
system on my whereabouts and adjusts my status accordingly. I usu- International Conference on Distributed Applications and Interopera-
ally have lunch around 12h30 and 13h00 together with my colleagues ble Systems (DAIS2005), Lecture Notes in Computer Science (LNCS).
in a room which is also used for meetings. Both time and place should Springer Verlag, (June 2005).
[9] Davy Preuveneers and Yolande Berbers, ‘Automated context-driven
match in order for my client to change to the ‘out for lunch’ status.
composition of pervasive services to alleviate non-functional concerns’,
If only the location matches, then my status should be ‘in a meet- International Journal of Computing and Information Sciences, 3(2),
ing’. Otherwise, if I am not in my office, I will ‘be right back’. Both 19–28, (August 2005).
location and time are modeled as fuzzy sets. [10] Thomas Strang, Claudia Linnhoff-Popien, and Korbinian Frank,
This simple test case with multiple fuzzy sets being matched ‘CoOL: A Context Ontology Language to enable Contextual Interop-
erability’, in LNCS 2893: Proceedings of 4th IFIP WG 6.1 Interna-
worked fine in 4 out of 5 cases. On one day I had lunch at 13h30, tional Conference on Distributed Applications and Interoperable Sys-
but had a meeting before at the same place. The instant messaging tems (DAIS2003), volume 2893 of Lecture Notes in Computer Science
client decided too early that I was out for lunch, and claimed that I (LNCS), pp. 236–247, Paris/France, (November 2003). Springer Verlag.
had a meeting while I was still having lunch. This was due to the fact [11] T. Buchholz, A. Kupper and M. Schiffers, ‘Quality of context: What
it is and why we need it’, in Proceedings of the 10th Workshop of
that the precision for the lunch time was set to high in order to match.
the OpenView University Association: OVUA’03, Geneva, Switzerland,
In the end, this simple approach using fuzzy set matching worked (July 2003).
rather well for this particular application. However, for a large num- [12] Mark Weiser, ‘The Computer for the Twenty-First Century’, Scientific
ber of fuzzy sets that have to match at the same time, it becomes very American, 99–104, (September 1991).
difficult to decide which context information matches best as more [13] Lotfi A. Zadeh, ‘Fuzzy sets’, Information and Control, (8), 338–383,
(1965).
and more scenarios will become equally likely.