New Approaches for Mainstreaming Metadata in Digital Library Project Development and Management
MetadataShow full item record
Metadata reflects a traditional notion of metadata as common descriptive elements abstracted from homogeneous collections of resources. Following this simplistic model, bibliographic terms, i.e. metadata elements, such as author, title, date, and subject can describe an collection or resource in that collection using traditional cataloging methods. Metadata have often been created to support discrete functionalities such as online discovery of large offline collections. In a world where many new library projects bring 100% of content online and deliver that content in diverse and dynamic formats, we must expand the role of metadata as a core resource for project development and management, not just for dissemination. Recent metadata standards such as METS (http://www.loc.gov/standards/mets/) provide more sophisticated descriptive elements but also reflect the increased structural and administrative metadata requirements of complex, open-ended projects consisting of complex objects. The Vivo virtual life sciences library (http://vivo.library.cornell.edu) includes no traditional body of content and publishes no metadata in any fixed format. Vivo serves as an index that cross-references people, departments, laboratories, equipment, publications, and events that collectively comprise the Cornell Life Sciences Initiative (http://lifesciences.cornell.edu/) and related educational and research activities on three campuses. Traditional metadata elements such as authorship, organizational affiliation, and subject keywords are subsumed into ontological <object>:<relationship>:<object> triples structured for the website database following the principles of the Harmony Project's ABC Ontology (http://metadata.net/harmony/JODI_Final.pdf)and other contemporary ontology design and integration methodologies. Many of these ontological relationships can be mined on-the-fly to produce standard metadata elements such as Dublin Core title, author, creator, or date for any object indexed in the collection, and the collection can be expanded with new types of objects and new relationships at any time in the project life cycle. Vivo's treatment of all data relationships as metadata and of metadata as relationship triples rather than individual elements offers new possibilities for flexibility and extensibility in creating and managing data and metadata for digital libraries.
Chapter 3 of 8.
Cornell University Library
CIP (CU Library Iniatives in Publishing)
Previously Published As
Metadata and Digital Collections: A Festschrift in Honor of Tom Turner; Ithaca, NY; CIP (CU Library Iniatives in Publishing); 2008; 54-81.