This course will become read-only in the near future. Tell us at community.p2pu.org if that is a problem.

Crowdsourcing


Lear what is crowdsourcing

Nowadays computer are really powerful, however there are some problems where computers are really bad. For example, imagine that I show you a picture and I ask you this simple question: do you see a human in this photo? While a person will answer yes or no almost instantly, a computer could even not answer it (imagine there is a kid with a Bob Sponge costume).

Humans, we, me and you are really good in several areas. Transcribing hand written text, image classification, 3D views, etc.

Therefore, in the last years there is a trend in the scientific community where specific problems are tacked by means of humans knowledge, not algorithms. Why? Because in several cases humans can and do faster tasks with a greater success than any possible algorithm.

Just to give you some examples. One of the very first scientific projects using this approach was the Stardust@Home project. The idea was to allow anyone with Internet access to help in the search for the first samples of solid matter from outside the solar system.

Stardust@Home "re-created" the microscope experience within a web-browser where people were able to zoom in and out of different samples, looking for some "dust" from the outter space.

Another very popular example is the Galaxy Zoo project. Galaxy Zoo uses the gorgeous Hubble imagery of hundreds of thousands of galaxies drawn from NASA's Hubble Space Telescope archive. The project presents these galaxies in a very friendly user interace, so people can help to classify them according to their shapes — a task at which your brain is better than even the most advanced computer. One of the most interesting aspects of this project, is that if you are quick, you may even be the first person in history to see each of the galaxies you're asked to classify.

There are many more web based citizen science projects, however in most of the cases there is no common framework where citizens or other researchers could deploy their own crowdsourcing applications where a community will help with their area of expertise.

PyBossa tries to provide this framework and service via a simple API and web interface that should be easy to use by any user or researcher. Right now, PyBossa features a few applications where people with an Internet connection can classify images or geolocalize specific features in a map.

The goal of this first task is to participate as a volunteer in one of the projects that using right now PyBossa and post here some comments about why did you choose the application and what could be improved. You can try these two projects:

Task Discussion