Metadata Games has several implemented games and our recent March 2011 pilot study on the tagging system using a control and experimental group with 200 of the same images showed some surprisingly accurate results, generating over 4000 new image tags. We’ll be doing a more public test very soon; contact us through contact at tiltfactordotorg if you want to help us test!
And Now, a message from our latest team member:
Hey there! I’m Parker, the newest member of the Metadata Games team. I’m joining the project with generous funding from a Dartmouth Digital Studies Fellowship. My first project with Metadata Games is to spruce up the librarian / archivist- facing feature set (otherwise known as the Backend). Pat the Librarian’s workflow will look like this:
* Pat easily uploads a collection of images to the Metadata Games
website using our online tools.
* The Metadata Games website gives Pat links to versions of the
Metadata Games that only use one particular dataset, which can be distributed to the library patrons, local volunteers, or vast internet players.
* The Metadata Games website also gives Pat dead-easy API access to
the tags that are generated by people playing Metadata Games on the
images. This makes it easy for Pat to use these crowdsourced tags to
improve the search and metadata descriptions on the collection’s
website.
The other part of my fellowship work involves using web scraping
techniques to provide machine access to several sets of public domain
images published by federal agencies, such as the CDC. This will allow
us to use these image sets as a public demonstration of the features
above. We can even take the tags generated by the cloud and pit them
against the tags given by the CDC to what the crowd can find that the
CDC misses!