Conference PaperPDF Available

Pictograms and assessment categories as crisis communication language: Lessons from a field exercise with GDACSmobile

Authors:

Abstract and Figures

In this article we explore how pictograms and assessment categories used by crisis management organizations support the crisis communication with the affected population. In this field exercise simulating a flooding event the exemplary tool GDACSmobile was used to let volunteers report their observations to the crisis management center using a report in which they assigned the category they believed was most fitting to the situation they reported. Despite volunteers reporting difficulty in selecting a fitting category, their actual decisions were highly fitting the intended categories defined by the crisis managers. We learned that pictograms and categories have potential as a common language between crisis managers as well as the affected population supporting the process of an effective crisis communication.
Content may be subject to copyright.
Pictograms and Assessment Categories as Crisis
Communication Language
Lessons from a Field Exercise with GDACSmobile
Roelof P. van den Berg, Adam Widera, Sandra Lechtenberg, Michael Middelhoff, and Bernd Hellingrath
Department of Information Systems
University of Muenster
Muenster, Germany
{roelof.vandenberg, adam.widera}@wi.uni-muenster.de, sandralechtenberg@uni-muenster.de,
{michael.middelhoff, bernd.hellingrath}@wi.uni-muenster.de
AbstractIn this article we explore how pictograms and
assessment categories used by crisis management organizations
support crisis communication with the affected population. In this
field exercise simulating a flooding event, the exemplary tool
GDACSmobile was used to let volunteers report their observations
to the crisis management center using a report in which they
assigned the category they believed was most fitting to the situation
at hand. Despite volunteers reporting difficulty in selecting a
fitting category, their actual decisions fit well with the intended
categories defined by the crisis managers. From this, we have
learned that pictograms and categories have potential to become a
common language between crisis managers and the affected
population, thus supporting effective crisis communication.
Keywordscrisis management, interaction with citizens,
language, communication
I. INTRODUCTION
Unexpected and dynamic crisis events cause a lack of
information for all involved actors. The affected population
wants to become aware of what has happened where and with
what consequences. The information required in these situations
varies from very generic, like what are the reasons for the
particular emergency, to very specific questions, such as local
availability of water pumps in order to dry basements. In
addition to the affected population, the involved crisis
management organizations also try to gather all available
information as fast as possible so they can plan and execute an
effective immediate response. Certainly, there are other actors
such as donors, press or the commercial sector, however, in this
paper we focus on the management of information exchange
between the affected population and crisis managers, the two
main target groups.
The information exchange between these two actors is an
iterative process covering the entire crisis management life
cycle, however, intensity and urgency is at its peak during the
immediate response phase [13]. In this paper we investigate, in
a broad sense, how information and communication
technologies (ICT) can support crisis communication taking
place between the warning of events and the immediate response
to those events. The main question is, how information flows
between the affected population, i. e. citizens, and crisis
management organizations, i.e. the responding authorities, can
be improved.
The main reason to raise this question is that current
practices, even in developed countries, rely mostly on classical
communication channels, mainly the telephone. This practice
creates bottlenecks while neglecting the potentials of current
ICT. The core competence of crisis management organizations
is operational response and not “customer care solutions via call
centers”. Therefore, capacity is a very tight restriction. Besides,
voice-based situation reporting is very time consuming and
partly inaccurate when considering the analysis of a report (e.g.
localization or interpretation). However, the speed and accuracy
of the provided information determines the degree of
effectiveness of the relief operation. The socio-technical
developments of the past decades, e. g. the growing availability
of smart phones or mobile networks, promise to improve crisis
communication and thus, have an impact on the effectiveness of
disaster relief operations [9]. Often these potentials are discussed
in context of the utilization of social media channels. However,
due to various drawbacks in the context of trustworthiness of
tweets and the organizational structures of responding
authorities [1], we limit this discussion on dedicated mobile
applications for crisis communication.
The utilization of potential mobile disaster, emergency and
crisis management applications faces several challenges.
Comparing the availability of several solutions (see e. g.
MissionMode 2016) one can ask, why the dissemination of non-
classical communication channels is so low. Reflecting the
current practices of crisis management practitioners, we have
identified a twofold gap: (1) in crisis situations the affected
population primarily gathers information in order to anticipate
new circumstances, sharing information is secondary; (2) the
experiences of crisis managers with the quality of mobile field
reports by citizens is rather bad [10].
Both mentioned phenomena have several reasons for
occurring, such as the reliability of the required hardware (e. g.
battery runtime), the quality of observations (e. g. non-urgent
requests during time-pressing sequences) or potential
information overload (e. g. through opening bottlenecks). One
general and overarching reason can be identified as neglecting
the socio-technical systems of crisis management organizations
and human behavior. Hence, the implementation of new ICT has
to focus on more than just the technical dimension, but rather
needs to consider all the interrelated pieces of crisis
communication. In other words, the real challenge in supporting
the described crisis communication is not technical but rather
addressing human behaviors in a socio-technical context. In this
paper, we mainly elaborate the appropriateness of the applied
semantics between citizens and crisis managers. We introduce
and investigate categories and pictograms as a translator
enabling effective incentive management between the two main
actors.
In order to do so we first introduce the basics on incentives
and incentive management. We qualify intrinsic and immaterial
incentives as main sources of motivation and introduce
category- and pictogram-driven communication as promising
approaches to increase information exchange between citizens
and crisis managers. Taking this as a foundation we present an
experiment setup within the EU-project “Driving Innovation in
Crisis Management for European Resilience” (DRIVER, Grant
agreement no 607798) with a focus on “Interaction with
Citizens”. During this experiment we deployed the human-
centered designed tool GDACSmobile as a data gathering tool
for the affected population. We present the applied category- and
pictogram-setup for a simulated flooding scenario as well as the
experiment results regarding the applied crisis communication
approach. We finally discuss the results and close the paper with
an outlook and next steps.
II. BACKGROUND AND RELATED WORK
The question whether or not a citizen uses a mobile app to
report an observation during a crisis situation depends on many
interrelated aspects. The question as to how far such an
observation supports the decision making of crisis managers or
not is a different one. In order to answer these two questions, we
first elaborate on which incentives should be activated in order
to motivate citizens to share their observations. In accordance
with appropriate incentive management, we present two selected
approaches as to how Volunteered Geographic Information
could enhance the processing time and quality of information
exchange between crisis managers and the affected population.
Why is it so important to understand the incentives of
citizens and to share their observations from the field? A good
example is the “90-9-1-Rule”. The so called “participation
inequality” states that on a large scale, multi-user communities
and online social network users do not participate to a great
extent. The amount of so called “lurkers” is 90% while 9%
contribute from time to time and only 1% contribute on a regular
basis [8]. According to current data, Wikipedia was accessed in
April 2016 nearly 16 billion times, while only 78 thousand
editors were active. Acknowledging a disaster as a “social”
event, a similar phenomenon can be observed in crisis situations.
The affected population wants to get as much information as
possible from authorities or first responders in order to decrease
information lag caused by the disaster. At the same time, the
information flow from the affected population to crisis managers
is means-oriented: in case there is a clear request, like need for
medical assistance or technical support, the authorities will get
contacted via 911, but the information around the actual request
which could be very important for situational awareness
remains a side note. The main reason is that time is a huge
bottleneck for both actors. In order to address this issue, the first
starting point needs to be the behavior citizens. The according
question is, which incentives need to be addressed and how, in
order to stimulate information exchange of the perceived ground
truth to the crisis management organizations.
Literature commonly understands incentives as instruments
influencing the behavior of members of an organization or
community in order to adapt to the organization wide system of
objectives [6]. By creating incentives, certain desired modes of
behavior from individuals are promoted, enabling specific
situational conditions which in turn result in the activation of
individual motives. In this case, a motive denotes a time-
invariant psychological disposition, i.e. an isolated, not yet
activated incitement for the behavior.
Several types of incentives can be differentiated and
systemized according to different classification criteria. The
differentiation of intrinsic and extrinsic incentives with respect
to the source of an incentive is a commonly accepted
understanding [4, 12]. Behavior is considered intrinsically
motivated whenever it is not a means to an end but stem from
the behavior itself. In a crisis management context, the readiness
to help a neighbor could be an example of such an incentive. On
the contrary, extrinsic incentives are generated by the
individual’s environment and are largely independent from the
type of task fulfilled. Additionally, extrinsic incentives may be
classified by the incentive object, which is differentiated into
immaterial and material incentives. Immaterial incentives can be
social in nature (e.g. relationship with neighbors) or be founded
in the nature of a task (e. g. serving as a nurse). Material
incentives can be further differentiated into monetary (e.g.
bonuses) and non-monetary incentives (e.g. shelter). Monetary
incentives play a minor role in the context of relief operations,
while there might be many non-monetary incentives relevant for
disaster relief.
The entirety of the given situational conditions determining
the behavior of members of a community is considered an
incentive system. An incentive system in the broader sense
corresponds with the structure and heterogeneity of a
community [2]. An incentive system in a narrower sense may
consist of structural as well as personal instruments of
motivation. The mode of action of incentive systems can be
explained reflecting an adjusted version of the “Valence,
Instrumentality, Expectancy” approach (VIE) developed by
Vroom [18] and the associated expectancy-valence theories.
Although initially developed for a work context (organization
and employee), many parallels can be found in the crisis
management context (community represented by the crisis
management authorities and the citizens as community
members). Expectancy-valance approaches assume that
motivation and consequently behavioral patterns are driven by
the pursuit of objectives. Individuals strive both for the
objectives of the entire organization or community (first order
objectives from the crisis manager perspective) and for
individual objectives (second order objectives). When imparting
extrinsic incentives, the first order results stemming from the
realization of these objectives have a direct influence on the
second order results, i.e. the individual achievement of
objectives. For example: The communication of a power
blackout from a citizen to the crisis managers may cause a fast
recovery being “useful” for the citizen and the entire affected
community. Valence in this context denotes the anticipated
value to which an individual attributes a specific action result
[18]. Depending on whether the result is considered desirable or
not, the valence takes positive or negative values. Example: The
report of a power blackout in another part of town may lead to
the need for sharing crisis management resources, i.e. a negative
valence. The dependency between first and second order results
is termed instrumentality [19]. From the actor’s perspective, first
order results are a means to an end for the realization of the
second order results. In the discussed example, the community
represented by the crisis management organizations tries to best
allocate the resources to the affected community and thus strives
for full information. The instrumentality denotes the expected
dependency of the second order result from the first order result.
Apart from the value attributed to a specific result, the VIE
approach also takes into consideration the individual
expectancies concerning the contribution of effort input.
Example: the reporting of a power black-out is higher, if the
reporter (or citizen) is affected or not. These expectancies in
addition to the valences build individual motivation for
conducting certain actions. According to the VIE approach, the
effort input of an individual is high when it expects that the
contribution will yield results, which firstly, are important to the
organization or community, and secondly, due to the expected
instrumentality, show close relationships to the individually
aspired results from extrinsic incentives, and thirdly, exhibit
valences as high as possible [18]. The VIE approach was
significantly extended by Porter and Lawler (see [4, 11, 12] for
more information). Figure 1 illustrates the key incentives which
according to this approach, influence the actions of individuals.
valence of
incentive individual abilities
and motivation
effort achieve-
ments
intrinsic
incentive
extrinsic
incentive
satisfaction
14
369
7a
7b
subjective
expectations
2
role
perception
5
experienced grade
of fairness
8
Figure 1: Expectancy-valence approach according to Porter
and Lawler [11]
In the center of this approach, are the individual efforts and
the achieved action results, e. g. reporting of a flooded road
(effort) leading to road closure (achievement). Individuals
define their effort by quantifying the valence of the reward (e.g.
value of road closure) and the subjective expectancies (e.g.
benefits of a road closure) in a first step. Those anticipated
achievements are related to intrinsic and extrinsic incentives
leading to satisfaction of the individuals. Depending on the
degree of satisfaction (e.g. public announcement of a flooded
road) and the experienced grade of fairness, the overall incentive
system stimulates the individual behavior within the community.
When referring these findings to the underlying question on
how information flows between crisis managers and citizens can
be improved, the valence of rewards and citizens’ expectations
need to be emphasized. I.e. if crisis managers want to receive
valuable reports from the ground, they need to install an intuitive
and easy-to-use information exchange with the community. As
stated by various researchers, involvement of citizens via VGI
especially in the response phase of the crisis management
lifecycle is of high importance [5]. Acknowledging the
perceptions of community members, a common language
visualizing the interactions is a basic requirement for an
appropriate crisis communication environment.
One way to establish a common language is to use
“Volunteered Graphic Information” systems (VGI, see also
[17]) in combination with commonly used information
categories and according pictograms. In the meantime, many
different VGI tools are available, having different pros and cons
regarding particular tasks (MissionMode 2016). However, as
mentioned above, the main challenge here is less a technical but
rather an organizational one. According to the discussion on
incentives we identify a trustful and open solution as most
appropriate. However, it must be able to visualize the benefits
for the community, i.e. an easy- and fast-to-understand situation
overview including a connection to responding authorities. In
order to do so, the information should be structured and
visualized respecting the communities’ attributes (like age
distribution, language, technical affinity, etc.).
Within the humanitarian domain, the categorization of
assessment reports is a commonplace practice. Currently, there
are several approaches to assessment categories that can be
found. Taking other humanitarian actors into account besides
governmental bodies, such as non-governmental organizations
and Red Cross Societies, the assessment categories provided by
the United Nations Office for the Coordination of Humanitarian
Affairs (UN OCHA) seems to have the biggest acceptance in the
field [17]. In combination with the Sphere project regarding the
minimum standards in humanitarian response, community-
driven clusters can also be identified [14].
In the area of crisis communication via symbols, it first must be
mentioned that many different standards exist in the field.
Pictograms and emoticons have added value from their ability to
increase communication speed and prevent translation errors in
natural languages. The EU funded INDIGO project delivered an
overview on symbol sets available and suitable for European
crisis management solutions [7]. UN OCHA since 2008 has
developed and openly published a set of 500 humanitarian icons
for fast communication in crisis situations [15]. These icons
have been adopted by many governmental and non-
governmental organizations all over the world.
In III.D we describe the subset of categories and pictograms
used for communication with citizens within this field exercise.
III. GDACSMOBILE USE CASE
A. Experimental Setup
Part of the DRIVER experiment aimed at streamlining
“Interaction with Citizens” in such a way that crisis managers
can use information obtained from citizens [3]. In order to assess
the usefulness of information for crisis managers, we compare
the classification of reports obtained from volunteers in the
experiment with the classification of their respective ground-
truth objects defined by the simulated disaster scenario (e.g.
blocked road) and the intended categories (infrastructure
roads). For the interaction with citizens one applied tool was
GDACSmobile. GDACSmobile is a mobile application using a
client-server architecture allowing citizens to report crisis
relevant observations to local authorities. A web interface is
used to moderate and visualize the incoming reports for the crisis
managers. More details on GDACSmobile can be found on [3]
or at http://crisismanagement.ercis.org/activities.
B. Setting
In the experiment, a combined storm surge with an imminent
flooding event in the region of The Hague, coinciding with high
tide threatens to overrun the coastal defense. In a morning
session, the preparation phase was simulated and the affected
population was asked to report situations that might become
dangerous during a flooding event (e.g. low fences or bike
racks). In an afternoon session, the response phase of a flooding
event was simulated in which the affected population was asked
to report where help was needed.
Figure 2: The sector maps as used to assign volunteers to
sectors. The left image (sector 1-6) is the northern area
while the right image (sector A-F) describes the southern
area. Volunteers assembled at the location indicated by the
circle within sector 3.
In the proposed scenario, the initial flood predictions are
limited to the northern part of the experimental area (Figure 2).
The morning scenario is therefore limited to this part and split
into 6 sectors, whereas the afternoon scenario covers the entire
region of Wateringse Veld with a total of 12 sectors. The
affected population was represented by volunteers who were
recruited using a variety of local and online media. Volunteers
were grouped (2-5 persons per group) so that each sector on the
map was visited and each group had at least one mobile device
with GDACSmobile installed. In the morning 7 unique mobile
devices were registered, and in the afternoon 15 were.
C. Observables
In preparation for the experiment, sets of markers (Table 1)
representing reportable situations were distributed throughout
the experiment region. Whenever these were encountered by
users of GDACSmobile, they were to be reported using the
mobile application.
Markers with text-only or photo-only representations of
information on the situation in the morning or afternoon were
attached to lamp-posts. Mobile markers were attached to
backpacks which were handed out to the volunteers that
participated as part of the affected population. For these markers,
no distinction was made between morning and afternoon. Next
to the markers present in the area, a list of objects of interest was
given to the volunteers who were asked to report these whenever
they encountered them.
Each group of volunteers was given an envelope containing
a handout on using GDACSmobile, a short text on the scenario,
and a list of addresses for the Text-Markers in their assigned
sector. The volunteers were explicitly asked to visit at minimum
all of the addresses on their list during each session lasting 1.5
hours. Only during the morning sessions, were volunteers asked
to look for real-world markers such as bike racks and high trees
that might pose a risk during a flood.
Table 1: The different types of objects distinguished
during the experiment
Object-
type
Amount
Example
Morning
Example
Afternoon
Text-
marker
71
playground
two kids trapped
by water height
Photo-
marker
11
quiet street
street under water
Object-
type
Amount
Example
Real-
world
10
bike racks, high trees
Mobile
9
cat in tree, burglars
D. Categories
For assessing the usefulness of information for crisis
managers, we classified the observed objects using a subset of
UNDAC [16] and Sphere [14] [categories and their definitions.
This subset was defined based on the roles of different first
responder groups involved in crisis management and is shown
in Table 2. The main categories were further illustrated using the
related UN OCHA symbols.
The used categories are accessed in GDACSmobile using
one of four main categories which cover all of the final
subcategories. An estimation of usefulness is made by first
checking if the reported main category matches the classification
and then if the subcategory matches as well. This way, three
levels of usefulness are defined: no match, half match, and full
match.
Table 2 pictograms and (sub)categories used on
GDACSmobile during the exercise
Category
Subcategories
Safety
Evacuation
Shelter
Emergency
Health
Medical treatment
Food and water supply
Health needs
Infrastructure
Buildings
Communication
Power, Water and Gas
Roads and Bridges
Other areas
Rescue
Displaced people
Missing person
Non-Human
Unreliant Person
IV. FINDINGS
An important initial impression to highlight first is that,
compared to other tools deployed during the simulation,
GDACSmobile was perceived as a proper tool to report
observations to crisis managers (four points on a five points
Likert scale). In this section, we show the results of the reports
categorization, so that the usefulness of the categories as a
language can later be discussed.
Out of the 195 obtained reports, only those containing geo-
located actionable information were quantitatively analyzed.
We experienced a few non-useful reports from text-markers due
to missing location data. Objects in the categories real-world and
mobile received almost only useful reports. Figure 3 shows the
amount of actionable reports obtained per information source.
The non-useful reports in the labelled Other
Figure 3 Actionable reports per information source
A. Matching of obtained reports
Multiple volunteers reported after the experiment, that they
found it difficult to define a category for the reports they sent in.
However, the matching of reported categories to expert
categorization showed that the volunteers’ matchings were
generally good. In total, only small amount (2.6%) of no match
reports where seen. 81.8% were fully matched to expert
categorization and 15.6% of reports were half matched.
Figure 4 Quality of matching per information source
From the sources defined in Table 1, the photo-markers were
only been reported three times. In Figure 4, the occurrence and
match quality of the other reports are seen. The reports on Text-
only and Real-world markers were best categorized by
volunteers with a full match ratio of 89% and 84% respectively.
Mobile markers in particular have a large amount of half-
matches, where the main category was estimated correctly, but
the subcategory does not match expert opinion. After inspection
of these reports, it is clear that all half-match reports were
categorized as “medical treatment” where experts would place
them in the category “health needs”.
Figure 5 gives an initial insight into the categories for which
such a mismatch often occurs. Although the health category was
the most often to be falsely categorized in relative terms (33%),
the infrastructure category saw the most categorization
mismatches in absolute terms (24 times).
Figure 5 Quality of matching per reported category
Further insight into mismatched categorized is given by
Figure 6. The clear connection between “health needs” and
71
41 17
13
25
3
0
22
0%
20%
40%
60%
80%
100%
Text Real-world Mobile Other
Usefull Non-useful
64 35
9
10
66
82
1001
0%
20%
40%
60%
80%
100%
Text Real-world Mobile Other
Full Match Half Match No Match
20
16 108 15
0822 1
3020
0%
20%
40%
60%
80%
100%
Safety Health Infrastructure Rescue
Full Match Half Match No Match
“medical treatment” shows the reports on mobile markers which
were categorized differently from the intended categorization.
A second observation is that many reports categorized as
“buildings” had other intended categories. In V, we further
discuss how these occurring mismatches can be lessened.
Rescu e Infrastructure Hea lth Safet y
Pow er, Wa te r
and Gas
Buildings
Roads and
Bridg es
Communi-
cat io n
Ot her A reas
Emergency
She l ter
Evacuation
Medical
Treatment
Food and
Water Supply
Health Needs
Dis place d
Peo pl e
Missing
Person
Non-Human
Rescu e Infrastructure Hea lth Safet y
Pow er, Wa te r
and Gas
Buildings
Roads and
Bridg es
Communi-
cat io n
Ot her A reas
Emergency
She l ter
Evacuation
Medical
Treatment
Food and
Water Supply
Health Needs
Dis place d
Peo pl e
Missing
Person
Non-Human
Intended
Category
Repo rt ed
Category
Figure 6 Matching of intended category to how they were
reported. Only lines for half and no match are shown.
B. Content of obtained reports
Because we assumed that the A4-sized markers would be
difficult to find in a larger area, we provided the volunteers with
a set of addresses in their assigned sector where markers were
placed. In general, the volunteers were able to find the markers.
However, some markers did not survive the entire day or were
located at a place without an exact address. Some reports (14)
were obtained informing about markers not found at the given
location. In addition, on occasion buildings close to the marker
location were reported without feedback on the missing
markers.
In the afternoon session, some reports came in which
conflicted with the water levels obtained from HKV1, a Dutch
consultancy and research agency in water and safety. An
example is a case where two reports were obtained within 50
1 HKV Ground truth maps on water levels
http://waterviewer.nl/#PR2505_10|Viewer
meters from one another. In the first report, the text of a text-
marker was reported with a water height of 1 meter. In the other,
volunteers seemed unable to find said marker and took a picture
of a supermarket where they assumed that a useful supply of
food could be obtained. However, the water level at that
location would have made the supply less useful, but this
knowledge was unavailable at the location.
Some reports on text-markers contained extra information
in addition to the text on the markers. In some cases, this was
to emphasize the importance of some part of the information
(e.g. medical help needed). In other reports, pictures were added
to give a visual impression of the marker location. However,
most reports on text markers either had an image of the marker
itself or the literal text on it.
V. DISCUSSION
A. Crisis commmunication language
Although volunteers reported that they had difficulty
selecting the right category for reporting their observations,
analysis of the reports in IV shows that the majority of reports
were matched properly with only 2.6% of reports having no
main category matches at all. We therefore believe that
pictograms and assessment categories can be used by the
affected population to communicate with crisis managers.
Because these communication tools are also understood by the
crisis managers, we furthermore believe that pictograms and
assessment categories make for a common language between an
affected population and crisis managers.
Thus, we can conclude that using categories and
pictograms supports the crisis communication cycle. Authorities
can share information relevant to the community in an efficient
way, in order to meet the needs of the citizens. By doing so, crisis
managers became able to stimulate first order objectives like
announcing to not enter a particular area in a proper manner: the
community benefits by e.g. not blocking roads or other
bottlenecks. At the same time, the affected population receives
support in achieving their second order objectives, such as
finding the most appropriate roads or becoming aware of not
passable areas. This kind of positive relation becomes amplified
by decreasing the efforts of sharing new observations by the
population.
B. Mismatches in categorization
In reports that were half mismatched or not matched to the
intended category, two cases stood out. The first is found in the
reports that were labelled Medical Treatment but have Health
Needs as intended category. The half matched reports consisted
mainly of situations where medical treatment is needed (e.g.
injured people), which has an intended category “Health
Needs”. Medical Treatment as intended category covers those
situations where medical treatment can be found (e.g. doctor’s
office) whereas Health Needs are described as situations where
medical assistance is required (e.g. wounded persons). This
subcategorization is one of the few where a need and a resource
which can be linked to one another are shown as separate. We
believe that this mismatch is an effect of the freedom of
volunteers reporting both needs and resources in
GDACSmobile.
The second case which stands out in the results, is found in
the categorization of reports as Infrastructure. Since this
category is the largest in regard to the amount of reports as well
as in regard to the absolute amount of mismatches, the wide
variety of mismatches that were seen here was expected. On
inspection of these mismatched reports, we found that some
reports were actually difficult to place in one category and
therefore had “Other areas” as the intended category. Based on
these findings, we believe that the volunteers had difficulty
selecting the proper category due to the multitude of reportable
objects found in the reports.
C. Marker visibility
Given that some reports indicated a lack of photo-markers, and
most of these markers were retrieved post-exercise, we conclude
that the addresses given for the marker locations did not provide
a good enough description for the volunteers to always find the
markers. We believe that the visibility over distance of the A4-
markers was not good enough for the volunteers to guarantee
that they were found.
Next to the reported missing markers, we also saw a
significant difference between reported markers to which the
volunteers were guided (text-markers) compared to same size
markers without guidance (photo-markers). Due to the fact that
all markers were hung up in the affected area at random, but
volunteers were guided, they could have missed the photo-
markers. Therefore, the effect and need of guidance needs
further examination. We do however believe that guidance of
volunteers is recommended in field exercises where certain
events need to be reported.
VI. CONCLUSION
We believe that pictograms and categories have potential as
an appropriate language that can be understood by both crisis
managers as well as the affected population. Thus, pictogram
and category-based crisis communications support effective
incentive management. Even though some mismatches occurred
on a subcategory-level, the pictograms and main categories were
matched in 97,4% of the reports. 18,2% of the submitted reports
on the sub-category level had a different understanding than the
intended sub-category. We ascribe the mismatches on the
subcategory-level to the freedom that GDACSmobile gives to
mobile app users. Reports were given that mixed up needs and
resources in the categorization, or covered multiple events. In
summation, we can state that first order objectives the situation
awareness of the affected community and its authorities can
only be reached when the crisis managers language is
understood by the citizens. Of course, it still needs to be
investigated if and how citizens perceive the situation overview
of the tool, but what we can conclude based on the experiment
results, is that the language (in terms of pictograms and
categories) has a high potential to be used as a common crisis
communication language. These results could be used further to
address other effort-related costs of citizens during disasters in
order to stimulate contributions to community objectives
through the above described second order objectives, i.e.
individual interests. One example could be ping notifications
like “Are you ok?” or “Is area X flooded”. The outlook in section
VII covers our outlook on improvements to the categorization
system and GDACSmobile to better handle reports that have
caused miscategorization. All in all, we believe in the potential
of pictograms and categories for crisis communication efforts.
VII. OUTLOOK
Based on the discussion and conclusion we propose future work
which aims to improve the use of pictograms and categories for
crisis communication. We differentiate between improvements
to the GDACSmobile tool in A and to experiments in B.
A. Improvements to GDACSmobile
To help the users of GDACSmobile with the issues that we
encountered in their reports while using the designated
categories, we propose to (1) make a clearer distinction between
needs and resources in the categories, and (2) allow for the
application of multiple categories in the reports. By doing so, we
address the issues described in V.B.
B. Future experiments
As seen in IV.A the majority of reported categories were in the
Infrastructure category. To get a better understanding of the
usefulness of icons and categories as crisis communication
language we propose to have a balanced representation of
subcategories in future experiments.
In future experiments, we would like to test how the users
of GDACSmobile are met in their second order objectives. One
feature in the tool is the map functionality in which reports
accepted and published by crisis managers are shown. We want
to find out which functionalities can further support users in
their second order objectives and motivate them to contribute
their observations.
ACKNOWLEDGMENT
The research leading to these results has received funding from
the European Union Seventh Framework Programme
(FP7/2007- 2013) under grant agreement n° 607798. We thank
the DRIVER “Interaction with Citizens” experiment team that
has been working together for several months in order to
prepare, conduct and finally assess the field exercise in The
Hague. We would like to explicitly thank the crisis managers at
the Safety Region Haaglanden (http://www.vrh.nl/) and the
volunteers organized by the Dutch Red Cross present during the
days of the exercise. A special thanks goes out to Lex and Silvia
of Safety Region Haaglanden for hosting the field exercise,
their assistance in the marker design, and the coordination of
the many volunteers.
BIBLIOGRAPHY
[1] A. H. Tapia, K. Bajpai, B. J. Jansen, and J. Yen,
“Seeking the Trustworthy Tweet: Can Microblogged
Data Fit the Information Needs of Disaster Response
and Humanitarian Relief Organizations,” 2011. In
Proceedings of the 8th International Conference on
Information Systems for Crisis Response and
Management ISCRAM 2011: From early-warning
systems to preparedness and training, ed. ISCRAM.
[2] R. Chambers, Whose reality counts? Putting the first
last. London: Intermediate Technology, 1997.
[3] D. Havlik, J. Pielorz, and A. Widera, “Interaction with
Citizens Experiments: From Context-aware Alerting to
Crowdtasking,” [2016]. In ISCRAM 2016 Conference
Proceedings 13th International Conference on
Information Systems for Crisis Response and
Management.
[4] E. E. Lawler III and L. W. Porter, Effect of Performance
on Job Satisfaction, [1967] 1967.
[5] F. E. A. Horita, L. C. Degrossi, L. F. F. G. Assis, A.
Zipf, and J. Porto de Albuquerque, “The use of
Volunteered Geographic Information and
Crowdsourcing in Disaster Management: A Systematic
Literature Review,” [2013] In Proceedings of the
Nineteenth Americas Conference on Information
Systems.
[6] H. L. Grob, C. Buddendick, N. Albrecht, and A.
Widera, “Towards A Holistic Approach of E-learning
Incentive Management,” [2007] In Proceedings of the
IASK International Conference E-Activity and Leading
Technologies, 28094.
[7] H. Hermansson and L. NewloveEriksson, “Symbols,
Symbology and Systems: A Comprehensive Overview:
Deliverable Report of the EC-funded INDIGO Project
(GA No 242341),” 2011.
http://indigo.diginext.fr/EN/Documents/I4.4.3%20Symb
ols,%20Symbology%20and%20Systems.pdf (accessed
June 30, 2016).
[8] J. Nielsen, The 90-9-1 Rule for Participation Inequality
in Social Media and Online Communities, 2006.
https://www.nngroup.com/articles/participation-
inequality/ (accessed June 30, 2016).
[9] J. Crowley, “Disaster Relief 2.0: The Future of
Information Sharing in Humanitarian Emergencies,”
Washington, D.C., 2011.
[10] M. E. Keim, and E. K Noji, Emergent use of social
media: A new age of opportunity for disaster resilience.
Original article.
[11] E. E. Lawler, Pay and organization development.
Reading, Mass., 1981.
[12] L. W. Porter, and E. E. Lawler, Managerial attitudes
and performance. Homewood: Irwin, 1968.
[13] S. F. Schulz, Disaster Relief Logistics: Benefits of and
Impediments to Cooperation between Humanitarian
Organizations, Techn. Univ., Diss.--Berlin, 2009., 1. ed.
Kuehne Foundation book series on logistics 15. Berne:
Haupt, 2009.
http://www.gbv.de/dms/zbw/584840993.pdf.
[14] Sphere Project, Humanitarian charter and minimum
standards in humanitarian response, 3rd ed., 2011 ed.
[Geneva], Rugby: Sphere Project; Distributed by
Practical Action Pub., 2011.
[15] UN Office for the Coordination of Humanitarian
Affairs, OCHA launches 500 free humanitarian
symbols, 2012. http://www.unocha.org/top-stories/all-
stories/ocha-launches-500-free-humanitarian-symbols
(accessed June 30, 2016).
[16] United Nations, “United Nations Disaster Assessment
and Coordination: Field Handbook,” 2013.
https://docs.unocha.org/sites/dms/Documents/UNDAC
%20Handbook%202013_english_final.pdf.
[17] United Nations Office for the Coordination of
Humanitarian Affairs, Humanitarianism in the network
age: Including world humanitarian data and trends
2012. OCHA policy and studies series. S.l: United
Nations, 2013.
[18] Vroom, V. H., Work and motivation. New York:
Wiley, 1964.
... [33], Middelhoff et al. [32] and van den Berg et al. [22]): ...
... IV-Self-mobilization Consequently, by acknowledging the perceptions of community members, a common language visualizing interactions becomes a basic requirement for an appropriate crisis communication environment. One way to establish a common language is to use VGI systems in combination with commonly used information categories and according pictograms [22]. In the meantime, many different VGI tools are available, having different pros and cons regarding particular tasks. ...
Conference Paper
Full-text available
Volunteered geographic information (VGI) has been seen as useful information in times of disasters. Several authors have shown that VGI is useful for coping with preparedness and response phases of disaster management. However, because it is still a young technology, the use of VGI remains uncertain, due to its lack of strong reliability and validity. It is our assumption that to improve reliability and validity the promotion of citizen engagement (CE) is needed. CE is not new topic, but in the digital humanitarian context, it involves important factors that are not yet considered by disaster managers, such as communication processes, motivation of volunteers, different media for production of information, etc. To fill this gap, we identified a set of preliminary factors which should be considered to promote the involvement of volunteers in disaster management. These factors were derived from critical review of CE literature and from an analysis of lessons learned from an experiment on interaction with citizens carried out in context of the EU-project “DRIVER – Driving Innovation in Crisis Management for European Resilience”.
... [33], Middelhoff et al. [32] and van den Berg et al. [22]): ...
... Consequently, by acknowledging the perceptions of community members, a common language visualizing interactions becomes a basic requirement for an appropriate crisis communication environment. One way to establish a common language is to use VGI systems in combination with commonly used information categories and according pictograms [22]. In the meantime, many different VGI tools are available, having different pros and cons regarding particular tasks. ...
Conference Paper
Full-text available
Volunteered geographic information (VGI) has been seen as useful information in times of disasters. Several authors have shown that VGI is useful for coping with preparedness and response phases of disaster management. However, be-cause it is still a young technology, the use of VGI remains uncertain, due to its lack of strong reliability and validity. It is our assumption that to improve reliabil-ity and validity the promotion of citizen engagement (CE) is needed. CE is not new topic, but in the digital humanitarian context, it involves important factors that are not yet considered by disaster managers, such as communication process-es, motivation of volunteers, different media for production of information, etc. To fill this gap, we identified a set of preliminary factors which should be consid-ered to promote the involvement of volunteers in disaster management. These fac-tors were derived from critical review of CE literature and from an analysis of lessons learned from an experiment on interaction with citizens carried out in con-text of the EU-project “DRIVER – Driving Innovation in Crisis Management for European Resilience”.
Conference Paper
Full-text available
Crisis management (CM) organizations regularly face the challenge to assess the potential impact of a change in their socio-technical setup. No matter if a new software, a new tool, a simple workflow or a broader organizational structure become available, CM organizations need to estimate the potential added value under a high degree of uncertainty. In general, the more reliable information about the new solution is available, the more informed the decisions are. One promising way in assessing the potential impact of new CM solutions can be found through its application in an as realistic as possible and an as secure as necessary setup. However, such artificial scenarios like simulation exercises hold the risk of measuring the performance of the solution itself rather than its contribution to the CM operation. In this paper we review the state of the art in measuring crisis management performance, discuss the results in the context of performance measurement in general and present a performance measurement approach supporting a structured assessment of innovative CM solutions applied within collaborative demonstration project.
Conference Paper
Full-text available
The EU FP7 project DRIVER is conducting a number of experiments to assess the feasibility of addressing known deficiencies in crisis management. In this paper, we introduce experiments that investigate two-way communication solutions between crisis managers and citizens or unaffiliated volunteers. In the so-called " Interaction with Citizens " experiments we are testing the usability and acceptance of the various methods and tools that facilitate crisis communication at several levels. This includes: informing and alerting of citizens; micro-tasking of volunteers; gathering of situational information about the crisis incident from volunteers; and usage of this information to improve situation awareness. At the time of writing this paper, our " Interaction with Citizens " experiments are still under way. Therefore, this paper reports the lessons learned in the first two experiments along with the experimental setup and expectations for the final experiment.
Conference Paper
Full-text available
The number of crisis events around the world has been increasing in the last years and suggests there is a real need to make communities more resilient to them. In addition to providing conventional authoritative data, ordinary citizens and residents in the affected areas are also voluntarily supplying information about the affected areas, in what has been called Crowdsourced or Volunteered Geographic Information (VGI). This paper conducts a Systemic Literature Review aimed at assessing the current state of research in the use of VGI as a source of information to aid the management of disasters. The results suggest there is an increasing body of knowledge of VGI and the way it can improve disaster management. It also reveals gaps in the use of VGI in the research areas of 'preparedness' and 'recovery', as well as the need for more robust case studies and experimental research to support this promising field.
Thesis
Eine ständig wachsende Zahl an Naturkatastrophen erfordert eine effizientere und effektivere Planung und Durchführung von internationaler Katastrophenhilfe. Da die Logistikkosten einen Großteil der Ausgaben von Hilfsorganisationen für Katastrophenhilfe ausmachen, stellen sie einen Haupthebel für Kostensenkungen aber auch für Verbesserungen hinsichtlich Lieferzeiten und Qualitätsniveau dar. Obgleich Investitionen in die Supply Chain Infrastruktur und andere vorbereitende Maßnahmen im Vorfeld von Katastrophen eine hohe Rentabilität versprechen, finden sie weder systematisch noch umfassend statt. Fehlende Kooperation und Koordination zwischen Hilfsorganisationen zählt dabei zu den am häufigsten kritisierten Aspekten. Aber warum sollten und wie könnten Hilfsorganisationen hinsichtlich ihrer Logistik in der Katastrophenhilfe zusammenarbeiten? Die vorliegende Promotion untersucht welche Synergien und resultierenden Vorteile eine horizontale Logistikkooperation zwischen Hilfsorganisationen generieren könnte und welche Hürden hinsichtlich ihrer Realisierung bestehen. Der Fokus liegt dabei auf permanenten Supply Chain Strukturen und den logistischen Hauptläufen bis zum "Point of Entry", sprich dem der Katastrophenregion am nächsten gelegenen Flug- oder Seehafen. Die Untersuchung konzentriert sich dabei auf ein Service Provider Modell als Art der horizontalen Zusammenarbeit, bei dem eine Hilfsorganisation als Logistik Service Provider für andere Hilfsorganisationen agiert, wodurch sich Volumen konsolidieren und Größenvorteile realisieren lassen. Im Rahmen einer Kombination aus deduktivem und abduktivem Forschungsvorgehen basiert diese Arbeit sowohl of Forschungsergebnissen bezüglich horizontaler Logistikkooperationen im privaten Sektor als auch Erkenntnissen der Neuen Institutionenökonomik sowie der Produktionskostentheorie. Eine komparative Analyse von drei verschiedenen Logistikkooperationen, die entweder von großen Hilfs- oder Spendenorganisationen initiiert sind, zeigt, dass im humanitären Sektor die gleichen Synergiepotentiale gesehen werden wie im privaten Sektor, dass jedoch noch nicht alle Vorteile realisiert werden konnten. Das liegt zum Teil darin begründet, dass humanitäre Organisationen sich offensichtlich mehr auf Lieferzeiten und Qualitätsaspekte konzentrieren als auf die Ausschöpfung möglicher Kostenvorteile. Insbesondere kleinere Organisationen könnten von dem Zugang zu einem breiten logistischen Servicespektrum profitieren. Die Arbeit identifiziert vier Haupthürden, die die Kooperationsbereitschaft von Hilfsorganisationen im Bereich Katastrophenlogistik beeinträchtigen: die Wahrnehmung von Logistik als eigene Kernkompetenz, kulturelle Unterschiede und gegenseitiges Misstrauen, fehlende Transparenz hinsichtlich potentieller und realisierter Vorteile, sowie unzureichende Lieferkapazitäten. Es werden Lösungsansätze zur Überwindung dieser Hürden entwickelt sowie Modellanpassungen diskutiert um identifizierte Schwachpunkte der Zusammenarbeit zu vermeiden. Insbesondere der Austausch von humanitären Service Providern durch kommerzielle Anbieter wird hierbei als interessanter Ansatz gesehen, der Raum für weitere Forschung bietet. Implementierungsrichtlinien weisen auf die Hauptaufgaben für die verschiedenen Stakeholdergruppen wie Spendenorganisationen, humanitäre Organisationen, kommerzielle Serviceanbieter sowie Medien hin, um Thematik und Umsetzung von horizontale Kooperation in der Katastrophenlogistik in der Zukunft weiter voran zu treiben.
Article
Background Social media (SM) are forms of information and communication technology disseminated through social interaction. SM rely upon peer-to-peer (P2P) networks that are collaborative, decentralized, and community-driven transforming people from content consumers into content producers. The role of SM in disaster management galvanized during the world response to the 2010 Haiti earthquake. (Pew 2010) During the immediate aftermath, much of what people around the world were learning about the earthquake originated from SM sources. (Nielsenwire 2010) During the first 2 weeks following the earthquake, “texting” mobile phone users donated over $25 million to the American Red Cross. (Sysomos 2010) Both public and private response agencies used Google Maps™. Millions joined MySpace™ and Facebook ™discussion groups to share information, donate money, and offer support. SM has also been described as “remarkably well organized, self correcting, accurate and concentrated”, calling into question the ingrained view of unidirectional, official-to-public information broadcasts. (Sutton, et al 2008) SM may also offer potential psychological benefit for vulnerable populations gained through participation as stakeholders in the response. (Sutton, et al 2008) (Laor 2003) Discussion However, widespread use of SM also involves several important challenges for disaster management. Although SM is growing rapidly, it remains less widespread and accessible than traditional media. Also, public officials often view person to person communications as “backchannels” with potential to spread misinformation and rumor. (Akre 2010) In addition, in absence of the normal checks and balances that regulate traditional media, privacy rights violations can occur as people use SM to describe personal events and circumstances. (Palen 2007)
Article
The abstract for this document is available on CSA Illumina.To view the Abstract, click the Abstract button above the document title.