Soylent: A Word Processor with a Crowd Inside -Setor


Soylent is a word processor crowdsourcing application that allows Amazon’s Mechanical Turks to improve, proofread documents on demand.  The application is broken into three main segments: shortn, crowdproof, and the human macro. Their contributions to this paper include using the paid crowd workers to analyze and understand tasks that the computer cannot do. The authors also explain the Find-Fix-Verify pattern that allows their application to work as well as analyze and evaluate the each component.

Amazing paper! As it gave me great insight into my own research. For one of the applications we are developing we are creating a video annotation tool to label and score videos of stroke survivors performing activities of daily living. Their methodologies of using 5 human macro scenarios showed promising results. I like the idea of using open source as a method of helping proof read and simplify papers. I am curious to know more background and research on the Mechanical Turks who are performing the crowd sourcing.Imagine how this application can be used in other domains or how it can be integrated to existing technology such as Microsoft Word. I am imaging using this technology on Microsoft Word and can only imagine how much easier to would make writing research papers! I can see this application developing further as well as becoming generalizable to applications as it is very useful for labeling data and proving great insight into the assistive technology domain.