12
edits
Changes
no edit summary
The Book recommendations database stores book recommendations written by library staff from all over Norway and links them to the RDF-representation of the MARC-catalogue.
==Economics of Scale: Thinking about Metadata Quality and Completeness for Fun and Profit==
* William Hicks, University of North Texas (William.hicks@unt.edu)
The UNT Libraries Digital Collections constitute three internet gateways, The Portal to Texas History, UNT Digital Library, and the Gateway to Oklahoma History, making available to the public a wide range of materials, from photographs and newspapers, to dissertations and recordings of music ensemble performances. The collections disseminate over 500,000 unique items, that were used over 9 millions times last year and with growth trends in both areas shows no signs of slowing.
As the size and scope of our collections has grown, so to has a pressing need to think clearly about the quality of our metadata, the completeness of our records, and the most efficient way of doing metadata entry. Not surprisingly there have been a few things written on the subject and so over the last few months we’ve started writing new code and getting the infrastructure of our metadata editing system to a place where we can begin to test these ideas on our ever expanding dataset. What kinds of questions are we looking to answer, and what types of tools are we building? That’s what this talk will be all about, but here are a few ideas to ponder:
* What kinds of tools have we built, or can we employ to standardize data entry and aid the user in their input needs?
* How close does a metadata record come to a “completeness” standard? What does that even look like? What are the implications when we look at such a standard at scale?
* If we can identify what we think a “quality” metadata record “is”, historically speaking, how close do we get to that ideal?
* Does an item’s history matter? Can we quantify it and locate value in change through time?
* What are the economic costs of metadata entry? If we have enough quantifiable measures about the types of objects in our systems, and we can profile our data entry personnel, what can this say about optimizing staff time and return on investment?
* What sort of priorities are we setting for ourselves when we treat all items as equal, when clearly some types of materials get vastly more use by the public.
* Finally what kinds of analysis tools might we develop to gauge our overall metadata “health,” to steer projects, or to ultimately improve our systems for our end user’s needs?
Most of our questions are still quite open ended, and honestly we are just getting started down this road. But as digital collections grow, and library budgets realign or shrink, it becomes increasingly important to back up our assertions and opinions with numbers, and find more efficient ways to work with the resources we have.
[[:Category:Code4Lib2014]]
