A Different Kind of Controlled Vocabulary

I read Adam's post about his difficult element, Relation, and it also made me think about some of the other elements that require descriptions that don't necessarily have a straightforward controlled vocabulary to use.  How do you achieve consistency of description while still retaining the freedom that allows those less-controlled elements to serve as a useful complement to the more-controlled elements?

For example, if Subject requires a CV that limits it to specific words and phrases like "Tackling" and "Quarterback sack," description might fill in the gaps with keyword-heavy phrases describing the play in more detail. The job of the Relation element would be to link the images of this play, perhaps from the beginning of an attempted tackle through another player's attempt to block and ending with a successful sack.

As long as all the image catalogers are using basically the same style for these elements, all will be well.  However, if one describer says "Auburn Right Tackle Smith launches himself" and another says "Tiger offensive lineman number 89 attempts sack" while a third says "Auburn player tries to stop Alabama players by using interpretive dance," then it's not as useful.  And I'm going to admit right up front that that third describer is probably me, so I'm concerned about holding up my part of the job when it comes to actually describing these images according to the rules!

That got me to thinking about newspaper and magazine writing, and it occurred to me that they use style guides.  So maybe there's a football writing style guide out there that could help all of us to be consistent yet creative and thorough in our descriptions?  I looked around and there are some possible choices.  Maybe the owners of the more descriptive elements would be kind to the football-challenged among us and consider making a style guide recommended practice for our digital library.  Here are a few examples, but there are probably more out there, or the owners of the appropriate elements might just list some suggested terminology and style:

• AP Super Bowl Style Guide
• Football Nation style guide
• A-Z Guide to Reuters Sports Style

A Different Kind of Schema

Tonight in class, Dr. MacCall talked about ways to automate, or partially automate, metadata entry in digital libraries. While highly sophisticated automation is beyond the scope of our class project, it's useful to be aware of the possibilities in case we need them later in our careers. Using our football image library, Dr. MacCall suggested as an example that perhaps instead of manually typing in a whole, possibly difficult-to-spell name (e.g "Baumhower") every time, the image cataloger could instead type in the player number (in this case, 73) and lookup software would provide the correct name.

Of course, it couldn't really be that simple, and the discussion quickly turned to all the ways it could go wrong (reassigned numbers, duplicate numbers, partial numbers, and so on), as well as the myriad rules for uniform numbers allowed on the field every game. I never realized that football uniform numbers were so complicated!  Furthermore, that complicated uniform number system would have to be written into the rules of the number-to-name matching software.  Thus numbering rules intended to reduce confusion on the actual football field would be re-purposed as a schema intended to reduce errors in a virtual library of football images. How wonderfully poetic to think that the rules and structure of the game itself could become an integral part of creating a football image library!

The Wayback Machine and the Cornell Web Lab

The D-Lib article "A Research Library Based on the Historical Collections of the Internet Archive," published in February 2006, describes an initiative at Cornell University to create a web archive for social science research in partnership with the Internet Archive.

While the Wayback Machine, described in this excellent January 2015 New Yorker article, archives as much of the web as it can (450 billion pages as of this writing), it is not indexed or readily searchable other than by URL or date.  For social scientists looking to do serious analysis of social trends, more manageable (though still enormous), indexed collections are needed. Cornell's project aimed to harvest collections of archived web pages--approximately 10 billion at the time of the article--and develop methods of automated indexing to make them useful to researchers.  Access would be through scripts or APIs. Designers envisioned researchers conducting projects to trace the development of ideas across the internet, follow the spread of rumors and news, and investigate the influence of social networks.

As William Arms, one of the developers of the Cornell project notes in this article from 2008, researchers were spending as much as 90% of their time simply obtaining and cleaning up their data. "The Web Lab's strategy," he explains, "is to copy a large portion of the Internet Archive's Web collection to Cornell, mount it on a powerful computer system, organize it so that researchers have great flexibility in how they use it, and provide tools and services that minimize the effort required to use the data in research."

The project exists today as the Cornell Web Lab, and though direct links to its site were returning errors as of this writing, a detailed description of current activities can be found on this faculty page, and the tools and services suite is still available through Sourceforge.