TY - GEN
T1 - Whom to Query? Spatially-Blind Participatory Crowdsensing under Budget Constraints
AU - ElSherief, Mai
AU - Raghavendra, Ramya
AU - Vigil-Hayes, Morgan
AU - Belding, Elizabeth
N1 - Funding Information:
This work was funded in part by the NSF Graduate Research Fellowship Program under Grant No. DGE-1144085, and in part by US Army Research laboratory and the UK Ministry of Defence under Agreement Number W911NF-15-R-0003. The authors would like to thank Hollaback for sharing their dataset.
PY - 2017/11/6
Y1 - 2017/11/6
N2 - The ubiquity of sensors has introduced a variety of new opportunities for data collection. In this paper, we attempt to answer the question: Given M workers in a spatial environment and N probing resources, where N < M, which N workers should be queried to answer a specific question? To solve this research question, we propose two querying algorithms: one that exploits worker feedback (DispNN) and one that does not rely on worker feedback (DispMax). We evaluate DispNN and DispMax algorithms on two different event distributions: clustered and complete spatial randomness. We then apply the algorithms to a dataset of actual street harassment events provided by Hollaback. The proposed algorithms outperform a random selection approach by up to 30%, a random selection approach with feedback by up to 35%, a greedy heuristic by up to 5x times, and cover up to a median of 96% of the incidents.
AB - The ubiquity of sensors has introduced a variety of new opportunities for data collection. In this paper, we attempt to answer the question: Given M workers in a spatial environment and N probing resources, where N < M, which N workers should be queried to answer a specific question? To solve this research question, we propose two querying algorithms: one that exploits worker feedback (DispNN) and one that does not rely on worker feedback (DispMax). We evaluate DispNN and DispMax algorithms on two different event distributions: clustered and complete spatial randomness. We then apply the algorithms to a dataset of actual street harassment events provided by Hollaback. The proposed algorithms outperform a random selection approach by up to 30%, a random selection approach with feedback by up to 35%, a greedy heuristic by up to 5x times, and cover up to a median of 96% of the incidents.
KW - Budget constraints
KW - Event detection
KW - Participatory crowdsensing
UR - http://www.scopus.com/inward/record.url?scp=85041392413&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85041392413&partnerID=8YFLogxK
U2 - 10.1145/3139243.3139249
DO - 10.1145/3139243.3139249
M3 - Conference contribution
AN - SCOPUS:85041392413
T3 - CrowdSenSys 2017 - Proceedings of the 1st ACM Workshop on Mobile Crowdsensing Systems and Applications, Part of SenSys 2017
SP - 31
EP - 37
BT - CrowdSenSys 2017 - Proceedings of the 1st ACM Workshop on Mobile Crowdsensing Systems and Applications, Part of SenSys 2017
A2 - Eskicioglu, Rasit
PB - Association for Computing Machinery, Inc
T2 - 1st ACM Workshop on Mobile Crowdsensing Systems and Applications, CrowdSenSys 2017
Y2 - 5 November 2017
ER -