Limitations of Traditional Training Methods for Crowd Workers in Machine Classification and Labeling Tasks

Meer over het boek

Training crowd workers before assigning them to tasks addresses the challenge of ensuring adequate domain knowledge. This method allows for the exclusion of underperforming workers early in the process. However, relying solely on a fixed training set and the number of correctly answered questions may not effectively determine a worker's suitability. The text argues for a more nuanced approach that considers each individual's learning pattern, suggesting that this could lead to better outcomes in the selection of crowd workers.