I've been thinking about two highly related (but still different tasks):
Creating a model that estimates the likelihood of correctness for any given label based on, for example, it's location on a street segment, it's x,y,z location on the labeling canvas, other labels on proximal streets, and possibly even external factors (e.g., census tract data, etc.)
Creating a model that estimates the quality of a crowd worker--some of which may take advantage of the first point but other features may be independent.
I've been thinking about two highly related (but still different tasks):
Creating a model that estimates the likelihood of correctness for any given label based on, for example, it's location on a street segment, it's x,y,z location on the labeling canvas, other labels on proximal streets, and possibly even external factors (e.g., census tract data, etc.)
Creating a model that estimates the quality of a crowd worker--some of which may take advantage of the first point but other features may be independent.