Citizen Sensing is a powerful paradigm involving citizens collectively participating in data collection. The pervasiveness of mobile devices has taken citizen sensing to unprecedented levels of adoption, as anyone with a phone can easily participate. However, especially when the collected data must be processed and analyzed by domain experts (such as municipal authorities), the flow of records is only useful if it does not exceed the receiving entity's capacity to process it. In this paper, we therefore envision and discuss the use of 2-Tier Citizen Sensing applications to mitigate this problem: In a first step (tier 1), data is collected by citizens in a crowdsourced fashion. In a second step (tier 2) the collected data is pre-processed by the crowd in a collaborative environment with aid of data mining algorithms, e.g. by aggregating redundant submitted records. The resulting reduced data can then be faster processed by domain experts. We evaluate this approach on data from an existing prototype of a tier-2 participatory urban infrastructure monitoring platform for civic issue reporting. By applying sentiment analysis algorithms on a large corpus of experience reviews (from 282 crowd-workers) we found relevant correlations between users' review sentiment scores and their performance on the crowd-working tasks. We show how these findings can be exploited to design better crowdsourcing platforms and discuss how citizens, domain experts and municipal authorities can benefit from such two-tier citizen sensing systems.
[1]
Alok N. Choudhary,et al.
Sentiment Analysis of Conditional Sentences
,
2009,
EMNLP.
[2]
Sajal K. Das,et al.
Incentive Mechanisms for Participatory Sensing
,
2015,
ACM Trans. Sens. Networks.
[3]
Paul Brown,et al.
Fix my street or else: using the internet to voice local public service concerns
,
2007,
ICEGOV '07.
[4]
Matthias Budde,et al.
Enabling low-cost particulate matter measurement for participatory sensing scenarios
,
2013,
MUM.
[5]
Matthias Budde,et al.
EstaVis: A Real-World Interactive Platform for Crowdsourced Visual Urban Analytics
,
2016,
Urb-IoT.
[6]
Walter Daelemans,et al.
Pattern for Python
,
2012,
J. Mach. Learn. Res..
[7]
Ramachandran Ramjee,et al.
Nericell: rich monitoring of road and traffic conditions using mobile smartphones
,
2008,
SenSys '08.
[8]
Lee Anne Fennell,et al.
Crowdsourcing Land Use
,
2013
.
[9]
J. B. Brooke,et al.
SUS: A 'Quick and Dirty' Usability Scale
,
1996
.
[10]
Holger Ziekow,et al.
Municipal Benefits of Participatory Urban Sensing: A Simulation Approach and Case Validation
,
2012,
J. Theor. Appl. Electron. Commer. Res..
[11]
Matthias Budde,et al.
Leveraging spatio-temporal clustering for participatory urban infrastructure monitoring
,
2014,
Urb-IoT.
[12]
Mark H. Hansen,et al.
Participatory sensing - eScholarship
,
2006
.