Into the Unown: Improving location-based gamified crowdsourcing solutions for geo data gathering

Abstract

Crowdsourcing has been recently successfully used to gather rich geo data in location-based games such as Pokémon GO. However, the academic literature has demonstrated that the selected crowdsourcing practices have led to the reinforcement of existing geographic biases, favouring rich areas and urban neighbourhoods over poor and rural areas. In this work, we investigate through two studies whether these biases could be mitigated by improving the crowdsourcing platform (Study 1) and supporting the crowdsourcing tasks with open map resources (Study 2). As an outcome of the first study, we derived 15 recommendations across six thematic areas for optimising the crowdsourcing processes. In the follow-up study, we demonstrated with a proof-of-concept work the potential to computationally improve the point of interest coverage particularly in developing countries and rural areas, and highlighted the potential of utilising open map services to build decision support systems for assisting in the evaluation of the crowdsourced content.