2022 LD4 Conference on Linked Data

The transition to linked open data requires that complex legacy formats such as MARC 21 be mapped to linked open data ontologies and vocabularies. In practice, metadata crosswalks are typically documented using spreadsheets and implemented in a transformation language such as XSLT (Extensible Stylesheet Language Transformations). This approach hinders interoperability because mapping spreadsheets vary from institution to institution: they are ad hoc rather than standardized, and they provide limited contextual information to guide those who are ultimately responsible for implementing a crosswalk in code. Although there are standards for defining metadata application profiles, there are no standard frameworks to guide the specification of metadata crosswalks.

As part of a campus-wide cross-collection discovery initiative, the Yale University Library is transforming its MARC-based metadata to the Linked Art profile of the CIDOC-CRM ontology. The library’s MARC/Linked Art crosswalk has been specified using DITA (Darwin Information Typing Architecture), an XML standard for documentation and technical writing that emphasizes structure, modularity, and content reuse. The DITA standard is topic-based and supports specialization: documents that provide information to contextualize a metadata concept can be paired with documents that specify individual tasks or algorithms needed to implement a crosswalk.