=Paper=
{{Paper
|id=Vol-431/paper-25
|storemode=property
|title=Matching ontologies for emergency evacuation plans
|pdfUrl=https://ceur-ws.org/Vol-431/om2008_poster7.pdf
|volume=Vol-431
|dblpUrl=https://dblp.org/rec/conf/semweb/MionPMA08
}}
==Matching ontologies for emergency evacuation plans==
https://ceur-ws.org/Vol-431/om2008_poster7.pdf
Matching ontologies for emergency
evacuation plans
Luca Mion1 , Ivan Pilati1 , David Macii2 , and Fabio Andreatta3
1
TasLab – Informatica Trentina S.p.A.
{luca.mion,ivan.pilati}@infotn.it, http://www.taslab.eu
2
Dept. of Information Engineering and Computer Science, University of Trento
macii@disi.unitn.it
3
Dept. of Civil Protection, Autonomous Province of Trento
fabio.andreatta@provincia.tn.it
Abstract. In case of emergency, the coordination of different services
deals with different working methods, different languages, different in-
struments, different sensors and different data representations. Thus,
the coordination of services includes heterogeneity problems that can
be managed with the help of ontology matching techniques. In this pa-
per we present a scenario where the requirements for ontology matching
arise from emergency evacuation plans, in the specific domain of civil
protection applications. We envisage what kind of smart sensor tech-
nologies could be used to support critical decisions when heterogeneous
sources of information have to be matched.
1 Introduction
In the context of semantic web and web services, heterogeneity represents a key
feature. One of the critical issues of semantic web services is the way the resources
of the semantic web have to be integrated as a whole. In fact, the ontologies
that are used to express information by means of sets of discrete entities (e.g.,
classes, properties, rules) are affected by heterogeneity, which requires proper
integration techniques [1, 2]. Ontology matching, namely the ability of finding
suitable relationships between entities from different ontologies, is essential to
achieve semantic interoperability and it may have huge social impact.
For example, when a large–scale disaster occurs many people from different
organizations may reach the critical area in a short time, and the need for inte-
gration of heterogeneous and rapidly evolving sources of data emerges. In this
context disaster management is strongly related, at different levels of abstrac-
tion, to environmental monitoring and ambient computing [3]. From a practical
point of view, the monitoring of an area involved in a disaster can be regarded
as a special example of environmental monitoring. In general, environmental
monitoring applications require sensing different quantities (e.g., temperature,
moisture or brightness), possibly evolving in time and space, as well as some
information related to the physical context in which some services are available
[4]. The purpose of these services can be merely informative, or aimed at making
decisions. In situations where an impending danger may affect the life of several
people at the same time, the criticalness of decisions and the dynamics of all
Fig. 1. Agent communication, adapted from [6].
events must be analyzed effectively in real-time, and thus the evolution in time
and space of interesting quantities should be monitored as well.
In order to design flexible services in such extreme conditions, defining the
ontologies of the various smart devices employed (along with their capabilities)
and implementing suitable matching strategies is a viable and effective approach
[6]. Heterogeneous ontologies have to get in correspondence in order to under-
stand messages sent by related sensors. Sensors can perform ontology matching
by themselves or by taking advantage of alignment or matching services, and
when they find a mutual agreement they can transform the alignment in a pro-
gram that translates the messages in axioms enabling the interpretation of the
messages, as represented in Fig. 1.
This paper presents a potential real-world scenario where the ontology match-
ing requirements are related to the management of emergency evacuation plans
from large buildings. In particular, we envisage how management of emergency
evacuation plans from large buildings can take advantage from ontology match-
ing. The case study considered in the following, although still at a quite pre-
liminary stage, results from a close collaboration between the Civil Protection
Department, various research centers and local companies, and with the help of
both staff members and volunteers of various rescue corps.
The rest of the paper is organized as follows: Section 2 introduces the basic
issues related to emergency situations and specifically describes the requirements
of the emergency evacuation plan scenario posed to ontology matching. Then,
Section 3 summarizes the conclusions and outlines future work.
2 Matching ontologies in emergency situations
In case of emergency, an effective coordination of people from different orga-
nizations (e.g., civil protection, police, ambulance, fire brigades, red cross) is
essential. The services offered by such organizations are characterized by dif-
ferent working methods, different languages, different instruments, sensors, and
data representations. Thus, the coordination of services certainly includes prob-
lems requiring ontology matching. Several monitoring systems can indeed collect
data for different organisations and for different purposes, using different sensor
technologies.
Although the different solutions deal with huge amount of data, the interpre-
tation and analysis is not consistent. Proper standardisation of data collection
processes is necessary, including applied technology and data storing formats, to
facilitate communication between services on a more consistent basis. Also, un-
Fig. 2. Example of sensor communication and ontology matching based on sensor on-
tologies in order to facilitate decision making.
certainty over the network consistence arises during emergencies (e.g., in terms
of sensors full functionality, or possible interaction that may occur with robots
or automated agents with own sensors), as exemplified in Fig. 2.
2.1 Emergency evacuation plans
As presented in Sec. 1, applications evolve in changing environments where de-
vices are replaced and added, and then it is not possible to establish unique and
definitive ontologies. Thus, applications have to be expressed in terms of generic
features that are matched against the actual environment. An interesting civil
protection applicative scenario, in which smart sensor networks could be partic-
ularly useful, concerns with the optimal management of emergency evacuation
plans from large buildings. As known, all public buildings (e.g., offices, shopping
malls, schools) are usually equipped with a certain number of safety exits as well
as by evacuation plans that should be carefully respected, in particular in case
of dangerous events (e.g., fires or earthquakes). Usually, the basic safety require-
ments are defined by national regulations. However, the actual effectiveness of
any pre-set evacuation plan can be limited by several issues, such as the impos-
sibility for occasional visitors (e.g., customers) to know the evacuation strategy,
the unpredictability of crush behavior in panic conditions, the lack of informa-
tion about number and about the distribution of people inside the building. To
solve such problems a smart sensor network could be deployed in a building in
order to automate the whole evacuation process. This network might consist, for
instance of:
– Redundant smoke or gas sensors to detect the presence of fire or the risk of
an explosion (redundancy is essential in this case to reduce the risk of false
alarms);
– Low-cost micro-electro-mechanical accelerometers for seismic events mon-
itoring [8] placed along the most important architectural elements of the
building;
– Smart video people counters located in proximity to doors, stairs or corridors
in order to estimate in real–time not only the total amount of people in the
building, but also their distribution [9–11].
Fig. 3. Ontologies related to the objects found in the building during evacuation, in-
spired by [12], with instances linked to their classes.
The data collected by the various sensors could be transferred through wired or
wireless connections (e.g., Ethernet or Wifi) to a central server (suitably con-
nected to an emergency Uninterruptible Power Supply, UPS) which in turn could
activate visual or sound alarms in order to manage the evacuation process in the
safest possible way. For instance, when a fire is detected, a software application
running on the server could estimate the level of risk in each area of the building
and then switching on the emergency signals and the way-out light indicators,
keeping into account the position and the distance of different users from the im-
pending danger. In this way, people could be safely and orderly guided towards
the safest exits, and in addiction the intrinsic risks related to a mass evacuation
(especially for children, for elderly people and for people with disabilities) could
be significantly reduced.
2.2 Emerging requirements
Usability of devices depicted in Fig. 2 is unpredictable since they are subjected
to being added/replaced or malfunctioning at any time. For instance, when a
robot enters a building during evacuation, it will introduce sensors that will pro-
vide more precision or information which has not been considered at application
design time and again ontology description languages can help solving this prob-
lem [12]. Fig. 3 shows possible specific ontologies related to the objects found in
the building during evacuation.
From this scenario we can derive requirements for matching solutions in the
context of emergency evacuation plans. In particular, requirements concern spe-
cific behaviours, such as requirements of being automatic (not relying on user
feedback), being correct (not delivering incorrect matches), being complete (de-
livering all matches) and being performed at run time. These requirements con-
firm the application requirements reported in [6], with reference to multi-agent
communication. Another important requirement concerns the execution time,
which has been indicated to be under 2 seconds by the Civil Protection staff, in
order to operate under stable and safe conditions.
3 Conclusions
In this paper, an applicative example is proposed in which the joint application
of both ontology matching strategies and smart sensor networks can be suc-
cessfully used to optimize building evacuation. Multiple sensors could be used
to estimate in real–time the total amount of people and their distribution in
the building, while proper matching of the sensor ontologies should facilitate
and greatly improve the decision making process. Future works include studies
to elaborate and to formalize the scenario, to choose and to develop a suitable
matching algorithm (e.g., as in [13]), and and extensive end-to-end testing.
Acknowledgments. We acknowledge the Autonomous Province of Trento for
supporting TasLab Living Lab and the European Network of Living Labs [14]
for promoting innovation activities in ICT.
References
1. F. Giunchiglia, M. Yatskevich, P. Avesani, and P. Shvaiko, “A Large Scale Dataset
for the Evaluation of Ontology Matching Systems,” Knowledge Engineering Review
Journal, 2008.
2. J. Euzenat, “Alignment infrastructure for ontology mediation and other applica-
tions,” in Proc. 1st ICSOC Int. Workshop on Mediation in semantic web services,
pp. 81–95, 2005.
3. J. Carlos , A. Jun, and L. L. Chen, “Using Ambient Intelligence for Disaster Man-
agement,” in: Knowledge-Based Intelligent Information and Engineering Systems
(KES2006), pp. 171–178, 2006.
4. O. Khriyenko and V. Terziyan, “Context description framework for the semantic
web,” In Proc. Context 2005 Context representation and reasoning workshop, 2005.
5. TasLab Webpage. http://www.taslab.eu.
6. J. Euzenat and P. Shvaiko, “Ontology matching,” Springer, Heidelberg (DE), 2007.
7. M. Marchese, L. Vaccari, P. Shvaiko, and J. Pane, “An Application of Approximate
Ontology Matching in eResponse,” In Proc. of ISCRAM, 2008.
8. REF TEK Webpage, http://www.reftek.com.
9. A. Bevilacqua, L. Di Stefano, and P. Azzari, “People Tracking Using a Time-of-
Flight Depth Sensor,” IEEE Computer Society, 2006.
10. Neuricam Webpage. http://www.neuricam.com.
11. F. Leonardi and D. Macii, “An Uncertainty Model for People Counters based
on Video Sensors,” Proc. IEEE International Workshop on Advanced Methods for
Uncertainty Estimation in Measurement (AMUEM), pp. 62–66, 2008.
12. J. Euzenat, J. Pierson, and F. Ramparani, “Dynamic context management for
pervasive applications,” Knowledge engineering review, 23(1):21–49, 2008.
13. F. Giunchiglia, F. McNeill, M. Yatskevich, J. Pane, P. Besana, and P. Shvaiko,
“Approximate structure-preserving semantic matching,” In Proc. of ODBASE, 2008.
14. European Network of Living Labs Webpage. http://www.openlivinglabs.eu.