Metadata Games

Earlier this month, I attended the New England Archivists conference in Providence, Rhode Island.  I attended a fascinating session there that was worth the price of admission for the whole conference.  The session was called, "Playing with Metadata:  Developing an Open Source Metadata Games System for Archives and Libraries."  The speakers were Mary Flanagan, Professor in Digital Humanities at Dartmouth College; and Peter Carini, College Archivist at Dartmouth.  With help from grants from the National Endowment for the Humanities, they are experimenting with ways to get one of the most daunting digitization tasks - creating metadata - to be done for us for free by outsiders.  How do they propose doing this?  By creating fun (and possibly addicting) computer games where the ultimate goal is to create good metadata!

The testing of these games is in the very early stages.  They are still experimenting with issues like how to determine by an algorithm whether a term is accurate or not.  They want the computer to do as much evaluation of terms as possible because if staff people have to spend a lot of time reviewing terms, then it defeats the purpose of the games.  They have toyed with the idea of giving more weight to a term the more times it is suggested by players.  But then what about cases where a player might have specialized knowledge? For example, the College used photographs from a Polar Expedition collection for the initial test.  An Artic Studies student ended up being one of the testers, and he used terms such as umiak (a type of boat) when playing the game.  This specific term is even more accurate than "boat," or "kayak," but it is unlikely that a large number of game players drawn from the general public would come up with this term.  The project staff is working on a way to deal with this issue by getting the games to give points to players for obscure but meaningful words.

Like many of our libraries and archives, the Dartmouth Archives holds tens of thousands of historical photographs.  At the current staffing level, which is unlikely to change dramatically in the near future, it would take the Dartmouth staff generations to create metadata for all the images.  And that's assuming no new additions to the collection.  In instances where the staff has created metadata, the metadata is often vague or generic due to the heavy workload.  Terms like "campus" to describe images taken on the Dartmouth campus are not uncommon.  This lack of detailed metadata hinders access to these wonderful collections.  Material may be digitized and put online, but few people may actually find it without good metadata.

Even on campuses like Dartmouth, populated by high-achieving and erudite students, computer games are very popular.  Besides playing lots of computer games in their free time, students can register for courses like "Game Design Studio."  Once the faculty and staff thought of matching up archivists' pressing need for better metadata to improve access to digitized material with the public's obsession with computer games, it seemed like the perfect marriage.  If Dartmouth is successful with this endeavor, they are dedicated to providing these tools to archives throughout the country through Open Source Software.

For more information on this exciting project, including a chance to participate in further trials, visit the tiltfactor web site.  Tiltfactor is involved in several other game projects.  The archives-related project is called "Metadata Games."