Crowdsourcing of Data Annotations


Crowdsourcing Data Annotations simplifies the process of creating meaningful knowledge from urban data streams. The data labels are both crowd-sourced across a dynamic community of contributors, and gathered through autonomous and semi-supervised learning algorithms.

The first phase of this tool will allow experimenters and citizens to use pre-defined tag sets. Over time, the tool will develop to help experimenters retrieve annotations and data, add categories for tags, write applications to gather annotations, and ask citizens to supervise and “sense check” the predictions of the machine learning tool.

Through the Urban Data Observatory, citizens will be able to select data sources and assign tags. This would be similar to the systems used in projects like Openstreetmap to crowd-source an accurate map or in social networking apps where it is possible to tag places, events, or people on media. Let’s say you want to understand why there was an abrupt change in the pollution and sound levels from a certain sensor on a Sunday afternoon – the people who were there could tag a football match happening in that time and place.

Those interested in annotating or requesting simple annotation parameters won’t need any previous technical knowledge. To make more complex operations and integrate APIs into your experiment you will need to have an understanding of programmable API clients.

Both experimenters and citizens can use Crowdsourcing Data Annotations to create the foundations for a sustainable and scalable urban data knowledge base.

Back to tools

Sign up for updates from Organicity

We'll keep you up to date on events in your city, opportunities to get involved, progress on our Open Calls and citizen experiments.