EAGER: Reliable Data from Heterogeneous Groups of Citizen Scientists

New York University, New York NY

Investigators

Abstract

Citizen science involves the general public in research activities that are conducted in collaboration with professional scientists. Citizens' participation shortens the duration and lowers the costs of certain research activities. A key challenge inhibiting the widespread adoption of citizen science is guaranteeing the reliability of contributions submitted by volunteers. Traditional approaches have relied on redundant distribution of tasks, whereby multiple volunteers are indiscriminately assigned identical tasks. However, most citizen science projects suffer from a scarcity of long term contributors and an abundance of casual, short term volunteers. Drawing inspiration from species across every phylum of life where physical and behavioral heterogeneities are evolutionarily selected, this EArly-concept Grant for Exploratory Research (EAGER) project posits that heterogeneities in citizen scientists will improve the reliability of data gathered. The envisioned paradigm will promote the progress of science, by enabling researchers to quickly gather large quantities of reliable data with minimal changes to existing infrastructure. Outcomes of this project will be mutually beneficial to researchers and society at large: researchers will have more confidence in citizen science and put forward more exciting projects which will contrive to enhance the scientific literacy of the public. This research program seeks to demonstrate a novel methodology to cogently distribute tasks among volunteers based on prior performance, affinity to the project, and technical potential. Specifically, the project hypothesizes that data obtained from subsamples of participants that are highly heterogeneous in terms of individual attributes will lead to more reliable data, thereby enabling a significant reduction in the degree of task redundancy and an improvement in data quality. This hypothesis will be tested within Brooklyn Atlantis, an online citizen science project for monitoring the environmental health of the Gowanus Canal - a highly polluted Superfund site. In Brooklyn Atlantis, citizen scientists identify objects of interest in images taken from the surface of the canal through an aquatic robot. A series of studies will be performed to: i) elucidate the relationship between data reliability and individual attributes; ii) quantify the potential of data fusion to enhance quality and accuracy of contributions; and iii) understand the role of group heterogeneity on data reliability. Rigorous statistics and constrained optimization will drive the implementation of an optimal task allocation engine for use in distributed citizen science applications.