Repository logo
 

Design Strategies for Efficient Access to Mobile Device Users via Amazon Mechanical Turk

Accepted version
Peer-reviewed

Type

Conference Object

Change log

Authors

Jacques, JT 
Kristensson, PO 

Abstract

It is often challenging to access a pool of mobile device users and instruct them to perform an interactive task. Yet such data is often vital to provide design insight at various stages of the design process of a mobile application, service or system. We propose accessing a pool of mobile device users via the microtask market Amazon Mechanical Turk (MTurk). While mobile device users are still a minority on MTurk, they provide unique opportunities for requesters. Not only does catering to mobile device users expand the potential worker pool, but also offers access to realistic in situ mobile sensor data for a variety of pervasive computing applications. We present a design pattern that makes it easy to crowdsource tasks to mobile device users. We demonstrate this pattern to obtain sensor availability information for some of the most popular mobile device sensors, using HTML5 web APIs to access them. We find that there is a sizable pool of crowd workers on MTurk and we present two design strategies for efficiently accessing them. The first design strategy enables direct fast access to a small pool of mobile-only device users. The second strategy exploits our finding that a large number of desktop-based crowd workers are prepared to switch to their mobile device if prompted by a well-designed mobile task. Finally we discuss solution principles for sensor-dependent mobile systems and studies that will be underpinned by mobile device crowd workers.

Description

Keywords

crowdsourcing, web, mobile, devices, sensors

Journal Title

Proceedings of First ACM Workshop On Mobile Crowdsensing Systems And Applications

Conference Name

CrowdSense ’17

Journal ISSN

Volume Title

Publisher

Association of Computing Machinery
Sponsorship
EPSRC (1198)
EPSRC (via University of St Andrews) (11780)
Engineering and Physical Sciences Research Council (EP/N010558/1)
This work was supported by EPSRC grant EP/N010558/1.