Visualizing NLP annotations for Crowdsourcing
This addresses the scalability issue in crowdsourcing NLP annotations by enabling non-experts to produce reliable data, though it is incremental as it builds on existing visualization toolkits.
The paper tackles the problem of making NLP annotation accessible to non-expert crowdsourced workers by introducing CROWDANNO, a visualization toolkit that simplifies tasks like clustering and parsing with an interactive interface, resulting in high-quality labels as shown in user studies.
Visualizing NLP annotation is useful for the collection of training data for the statistical NLP approaches. Existing toolkits either provide limited visual aid, or introduce comprehensive operators to realize sophisticated linguistic rules. Workers must be well trained to use them. Their audience thus can hardly be scaled to large amounts of non-expert crowdsourced workers. In this paper, we present CROWDANNO, a visualization toolkit to allow crowd-sourced workers to annotate two general categories of NLP problems: clustering and parsing. Workers can finish the tasks with simplified operators in an interactive interface, and fix errors conveniently. User studies show our toolkit is very friendly to NLP non-experts, and allow them to produce high quality labels for several sophisticated problems. We release our source code and toolkit to spur future research.