The Cornell Lab of Ornithology’s eBird database contains more than 70 million records of birdwatching observations provided by scientists, researchers and amateur naturalists. Launched in 2002, eBird has been hailed as an example of democratizing science, creating a network of both recreational and professional bird watchers who can use their own data and the data generated by others on species distribution, migration timing, and habitat usage. The database is used by educators, land managers, ornithologists, and conservation biologists to better understand nature and the environment.
The project will focus on calculating summary statistics from the database, such as distribution of observations by the observer, choice of observation and locations of species observations. The statistics will then be used to attempt to understand the dynamics of observer behavior based on things like expertise and length of involvement in eBird. The overall goal of the project is to gain a better understanding of citizen scientists’ behavior and how to improve the quality of their observations through machine learning technology.
By incorporating human learning in a loop with machine learning to vet good and bad observations, the study aims to help citizen scientists become better birdwatchers and better trackers of quality observations. This will help to make citizen science data more valuable and help improve the quality of overall data.
This research is part of a larger grant distributed to Cornell, with Prinicpal Investigator Steve Kelling and Co-Principal Investigator Carla Gomes also collaborating.
To get Carl Lagoze's synopsis of the project, watch his YouTube interview here: