geometry, but creates a new version of it. All versions are kept 
indefinitely thereby adding a temporal dimension to the system. 
Propesttes Sensis onus 
Polygon psaperties 
  
Obstacle Route 
ir rr e Ter Th deem 
, e 
  
Figure l. The desktop interface. Polygons denote real-world 
obstacles. 
Figure 2 shows an overview of the system's components. 
Users’ observations are saved in the PostGIS database first and 
move later to GFT. GFT holds the latest snapshot of the data, 
whereas PostGIS stores all data ever entered. 
Analysis and cooperation: One of the presented application’s 
main merits is the stimulation of crowdsourced data collection. 
As already noted, the crowd is capable of more than data 
collection only. To leverage this capability, the built application 
supports the calculation of the shortest route. The routing 
functionality turns the application into more than a data silo. 
The purpose of this analysis is twofold. On one hand it 
provides rescue workers with an automated shortest path 
analysis tool. On the other hand it acts as a return of investment 
for the crowd as they can use their own data for their own 
routing needs. The people's usage of their own data acts as an 
incentive to generate better and accurate data, and keep it up to 
date. 
Users are encouraged to cooperate on several different levels. 
First, they can work together on collecting the most accurate 
information they can find through all the means they are 
comfortable with. For instance, some may draw information 
from their Twitter network, whereas others may have some 
experience working with satellite images. Both types of users 
can enter their data in the application and compare the results. A 
third user may then check both of their results. Another type of 
cooperation is found between the desktop and mobile users. The 
mobile users’ task is to make quick and numerous observations 
which the desktop user then synthesizes into polygons. Both 
users benefit: the mobile user can work autonomously without 
worrying about what other mobile users are doing, while the 
desktop user can observe the whole operation from a higher 
vantage point using more powerful hardware without needing to 
worry about the difficulties of working in the field. 
The tools available for collaboration are the comments section 
on the desktop and mobile interfaces and the communication 
facilities of GFT. GFT allows users to discuss all facets of the 
data in its tabular view. Users can place comments on columns 
rows and individual cells. 
(Social media 
| Information 
sources 
  
  
   
     
ER / Desktop | / Mobile | 
iinterface : Interface | 
Official t j rer 
sources | i Nd Dia 
  
   
AJAX, JSON 
    
bo Geometnes, um. 
Fs —-U| comments Bau 
GM Fusion | 
PostGIS Python API Tables | 
  
  
  
  
  
  
it id 
th |i 
En Ec £a i 
HE i 
Remote Hi il 
sensing | sus E 
products | i T r 
{ 60 : i © 5 X 
Developers General public 
Figure 2. The system's components. Data streams are represented 
by arrows. The users on top are the data collectors. They gather 
data and enter it through the desktop and mobile interfaces. The 
text on top and left of the lines denote the payload of the data 
stream while the text on the bottom and right of the lines denotes 
the used technology 
Data presentation, visualisation and sharing: as discussed in 
section 2.2, crowdsourcing initiatives are successful and gain 
momentum when they are open and tuned to the user's 
capabilities and needs. In terms of storage, open means that the 
data is easy to access. Two user groups are targeted here: 
"normal" users (informal part of second tier) who are not able or 
willing to work with raw data and prefer a pre-processed 
version of it, and savvy computer users (volunteers in the third 
tier) who want access to the raw data. The built application 
satisfies the needs for these two groups through the following 
means: the web and mobile interfaces, GFT's export 
functionalities and visualisations, and the developer API. The 
web and mobile interfaces allow people to not only input data, 
but to also browse the collected information. GFT's web- 
interface supports tabular and mapped visualisations. The data 
can be extracted as KML. The API gives access to the data in 
developer friendly formats such as Well-known Text, KML, 
GeoJSON, etc. 
5. CONCLUSION 
Crowdsourcing is the unplanned coming together of a highly 
diverse group of people who use the latest technology and data 
formats and sources available to aid a certain disaster 
management cause. It is difficult to predict beforehand how 
many people with what skills will participate and which tools 
they will deploy. Designing an application for a crowdsourcing 
effort therefore seems counter to its volatile nature. What 1s 
needed, rather, is not a complete system, but a set of 
components which are well developed, well documented and can 
450 
seamlessly bi 
needs. This v 
disaster mana 
The main str 
disaster mane 
analysis in th 
data for othe 
as they too c 
information i 
first-hand hoy 
the routing s 
might change 
“contribute cc 
communicatio! 
community . 
Parallels betw: 
initiatives and 
notice and fos 
technology th 
they have mo 
used technolo 
hackers to ada 
sourcing a pro 
turn results in 
The applicati 
deploying it ir 
and principles 
research enviro: 
Trust is an in 
Mechanisms 
trustworthiness 
implemented. 
brought forwai 
current app lice 
instance an ap] 
able to discus: 
obstacles. Imp 
strengthens th 
examination 
Àn in-depth s 
performed. Dra 
input method ax 
to assess the bes 
Currently, an ir 
to function. T| 
Cannot be taken 
Mechanism need 
to function in th 
Google Maps is 
the time of th