ECCO-A Framework for Ecological Data Collection and Management Involving Human Workers

Scientific and ecological data collection in today’s world is primarily driven by citizen-based observation networks to gather information on a diverse array of species and natural processes. Such efforts leverage the contributions of a broad recruitment of human observers to collect data and use Machine Learning algorithms to process the collected data leading to a computational power that far exceeds the sum of the individual parts. Instead of organic group formation and collaboration, our vision is the need to formalize collaboration and rethink the components of a data management system to ensure its sustainability in such human-intensive applications. The enabler of collaboration is the notion of a user group that implies different behaviors and interactions between its members. We advocate the design of new components of a data management system that deliberately acknowledge the uncertainty and dynamicity of human behavior by capturing the human factors that characterize group members. We describe ECCO, a framework that contains two generic components: adaptive collaborative human factors learning and adaptive human-centric optimization. Those are the core components that support the fundamental functionalities of a wide range of human-intensive applications. ECCO components rely on two optimization engines, namely task assignment and human data management engine. An additional challenge in designing the components of ECCO is the need to support adaptive and incremental computation. We discuss the modeling, learning, and computational challenges of designing the components of ECCO and propose a roadmap of future directions of this vision.