摘要: Crowdsourcing has become a powerful paradigm for accomplishing work quickly and at scale, but involves significant challenges in quality control. Researchers have developed algorithmic control approaches based on either worker outputs (such as gold standards or agreement) behavior task fingerprinting), each approach serious limitations, especially complex creative work. Human evaluation addresses these limitations does not scale well with increasing numbers of workers. We present CrowdScape, system that supports the human crowd through interactive visualization mixed initiative machine learning. The combines information about outputs, helping users to better understand harness crowd. describe discuss its utility grounded case studies. explore other contexts where CrowdScape's visualizations might be useful, such user