Optimization of Orchestration of Geocrowdsourcing Activities

Abstract

Abstract. Geocrowdsourcing has proved to be very valuable in crisis situations. Calling to citizen on the ground or to experts or trained amateurs to help in the mapping of crisis situation is a recognized and valuable practice. However, despite the experience gained from real and dramatic situations, it remains difficult to set up and execute complex processes that require actions of both people on the ground and people on the web, and to understand how to get the best result at the minimal cost in term of users actions. In this paper, we describe a process that can be used to assess a global situation on a map using a combination of services and user operations. We want to understand how best to distribute a limited amount of human actions between different kind of tasks in order to get the most reliable result. Since it is difficult to conduct experimentation, we have decided to use simulation to reach a result that could be applied on the ground. This simulation relies on a geolocalised corpus of tweets. It provides some hints about how to deploy an exercise on the ground that are discussed as a conclusion. In addition, we propose a binary integer programming (BIP) making best use of the available workers.