top of page
Crowdsourcing is a collaboration model that distributes tasks to a large pool of online contributors, typically coordinated through an online platform for parallel execution.
In data annotation, a dataset is split into many small units, often called microtasks or HITs, and assigned in parallel to many annotators to achieve fast and low-cost labeling at scale. Compared with professional outsourced annotation teams, annotation consistency and stability are often weaker.
Common platforms include Amazon Mechanical Turk, Appen, and Toloka, and typical quality controls include multi-annotator consensus, gold questions, and worker reputation scoring.

Outsourced data annotation project with anonymous data annotators

bottom of page


