Refine
Institute
Has Fulltext
- no (2)
Language
- English (2)
Document Type
Keywords
- behaviour modelling (1)
- interruptibility (1)
- mobile crowdsourcing (1)
- wearable computing (1)
Is part of the Bibliography
- no (2)
In daily life, we use computers and smartphones to organize a groups activities and share documents with its members. However, creating that group in the digital domain is often performed manually. Solutions to automate the process of forming a group of physically proximate members, so far, mostly relied on Bluetooth or WiFi. In contrast to these solutions we present Grouve, an automatic ad-hoc based group formation technique running based on ultrasonic waves. As Grouve uses inaudible sounds for communication, it has the natural advantage that the sound signal stays inside a limited space, which matches most of the application contexts. In this paper we report on the design process and implementation of Grouve.
Traditional urban-scale crowdsourcing approaches suffer from three caveats - lack of complete spatiotemporal coverage, lack of accurate information and lack of sustained engagement of crowd workers. In this paper, we argue that these caveats can be addressed by embedding crowdsourcing tasks into the daily routine of mobile workforces that roam around an urban area. As a use case, we take the bpost who deliver the letters and parcels to the citizens across entire Belgium. We present a study that explores the behavioural attributes of these mobile postal workers both quantitatively (6.3K) and qualitatively (6) to assess the opportunity of leveraging them for crowdsourcing tasks. We report their mobility pattern, workflow, and behavioural traits which collectively inform the design of a purpose-built crowdsourcing solution. In particular, our solution operates on two key techniques - route augmentation, and on-wearable interruptibility management. Together, these mechanisms enhance the spatial coverage, response accuracy and increase workers' engagement with crowdsourcing tasks. We describe these principal components in a wearable smartwatch application supported by a data management infrastructure. Finally, we report a first-of-its-kind real-world trial with ten postal workers for two weeks to assess the quality of road signs at the city centre of Antwerp. Our findings suggest that our solution was effective in achieving 89% spatial coverage and increasing response rate (83.6%) and accuracy (100%) of the crowdsourcing tasks. Although limited in scale, these and the rest of our findings highlight the way of building an efficient and purposeful crowdsourcing solution of the future.
