LibraryCloud for DPLA
What is LibraryCloud?
LibraryCloud is an open metadata server. It gathers metadata from libraries and other curating institutions and makes it available to developers and other sites through open APIs and as Linked Open Data.
What is it for?
The aim is to encourage and enable innovation using everything that libraries know. Apps that could be written using LibraryCloud's metadata include:
Browsing and discovery tools
Recommendation engines
Semantic webs of library materials
Social networks based around library resources
Library analytics — for local libraries and across multiple libraries
Open-access content delivery systems based on profiles, recommendations, and availability
But, of course, the real aim is to enable completely unexpected, disruptive innovation.
Which metadata?
LibraryCloud at this time is making the following data and metadata available (to a controlled set of alpha users, at this point):
12M bibliographic records from Harvard's 73 libraries
10 years of circulation records from Harvard, broken down by student, graduate student, and faculty member
10 years of records of recalled items from Harvard
Harvard data about books placed on reserve, used in courses, etc.
1M bibliographic records from San Francisco Public Library system
2.5M circulation records from San Francisco Public Library system
361,000 bibliographic records from the San Jose Public Library, with their associated circulation records
110,000 bibliographic and 146,000 circulation records from the Darien CT Public Library
Pointers to more than 400,000 Open Library works that are readable online, with links back into library catalog data about them
Links to 60,000 Wikipedia pages about books, about 10,000 NPR audio programs, 2,600 open courseware pages, etc.
More is on the way, including social usage data from LibraryThing...
Architecture
LibraryCloud is built on a LAMP stack. It keeps its bibliographic records in a mySQL database. These are put through SOLR for search and faceting.
LibraryCloud's APIs are available to anyone with an API key (required at least while it is in beta). The APIs are documented on the LibraryCloud wiki.
Take it for A sPIn
To try out the LibraryCloud API Query Builder, click the button:
The "trends page in ShelfLife is a quick example of what can be done with LibraryCloud. It does a simple API call for each of the five libraries, getting a JSON object in return.
LibraryCloud supports Linked Open Data
LibraryCloud's bibliographic data is available as Linked Open Data. (Because of various licenses and contracts, it is only available in small chunks, and is a subset of the complete bibliographic records.)
Here's a 9 minute interview with Matt Phillips, one of LibraryCloud's developers, explaining how LibraryCloud has been made Linked Open Data compliant:
Openness and Privacy
LibraryCloud's policy is to accept only data that can be made fully open and public, without license or encumbrance.
LibraryCloud adheres to a strict privacy policy worked out with the help of Harvard's chief privacy officer, Scott Bradner. This policy prohibits LibraryCloud from accepting data that has any user ID's or even encrypted or random ID's that would enable a maelfactor to piece together patterns of usage that might lead to identifying the user. In addition, LibraryCloud's recommends the contributing libraries not provide circulation data until after it has expunged the user IDs from its own records.
Data is time-stamped with the date, but not with any finer detail.
The question of what constitutes a proper degree of privacy protection is, of course, one for DPLA to decide; we would look forward to participating in that discussion.
LibraryCloud and the DPLA
We believe that LibraryCloud can enable DPLA to meet one of its most important goals: Enabling everything that libraries know to be used and re-used all across our increasingly networked global, national, regional, and local cultures.
Back to the launch page
The demo FAQ
The ShelfLife Collaborative
If you have questions, including technical support issues, please contact:
David Weinberger self@evident.com 617-852-6902
Kim Dulin kdulin@law.harvard.edu 617 496 3292
Paul Deschner deschner@law.harvard.edu 617 384 9799