Table of Contents
Introduction
One of the biggest hurdles to a successful repository migration (of any kind) is metadata: do you keep your metadata records in their current format or transform them to something new? What are the pros and cons of XML vs. RDF? Should you clean up your metadata prior to a migration? This document seeks to provide a framework to guide decision making in these areas.
Deciding How to Proceed
Fedora 6.x is capable of storing the same metadata file formats (e.g. XML) as Fedora 3.x, so mapping and transforming XML metadata fields to RDF as part of a migration is not strictly required. However, depending on your requirements, such a process may be beneficial or even required. Much depends on the expectations of your front-end environment.
Islandora
Islandora 2.0 uses Drupal fields for metadata, which are atomic. These fields map cleanly to RDF, but they lack the structure of an XML document. Therefore, if you plan to migrate to Islandora 2.0, you will need to map and transform your metadata to fields. Typically, this would be achieved by mapping each XML element to a field based on an RDF schema. Please see the section on mapping from XML to RDF for more information.
Samvera
Most Samvera applications (e.g. Hyrax, Hyku, Avalon) expect field-based RDF metadata for search and display. Therefore, if you are migrating to a Samvera application, you will likely need to map and transform your XML metadata elements to RDF fields.
Custom
If you have a custom Fedora 3.x interface that you plan to update as part of your migration to Fedora 6.x you will have some things to consider. If your interface supported XML metadata in Fedora 3.x you have the option of simply maintaining this support in the Fedora 6.x context. This is the most straight-forward option; you can use migration-utils to migrate all your Fedora 3.x content without worrying about mapping or transforming your metadata. However, if you want to change to RDF metadata now or in the future you will need to update your interface to support this new format. The pros and cons of XML vs. RDF metadata are described in the following section.
XML vs. RDF
XML has been the most common format for repository metadata for many years. Common descriptive schemas include Dublin Core and MODS. XML is highly structured, and schemas like MODS take advantage of this by using nested, hierarchical elements to express particular concepts. For example, the mods:titleInfo element contains several sub elements the provide information related to a title:
...
Ultimately, there are pros and cons to both RDF and XML. If you’re unsure how to proceed, it might be helpful to develop a set of functional requirements related to your metadata needs. These should include both the constraints imposed by your front-end application and the needs of your users.
Mapping from XML to RDF
If you decide to migrate to RDF metadata you’ll need to begin by documenting all your currently used metadata schemas and fields. This can be done using a combination of spreadsheets and documents; this MODS to RDF Mapping Template based on the Whitman College migration pilot work is a good starting point.
Community RDF Mapping Resources
The following community resources should be consulted regarding specific mapping advice.
- MODS to RDF Mapping Recommendations - See Samvera Metadata Working Group resources (Note: the MODS and RDF Descriptive Metadata Subgroup met all of their chartered deliverables and is not currently active)
- ARMINDA MODS to RDF Mappings
- Islandora MIG MODS-RDF Simplified Mapping - See current Islandora MIG documentation resources
- Example CLAW for MODS Mapping - See current Islandora MIG documentation resources
- Draft Recommendations See current Islandora MIG documentation resources
RDF Schemas
The following RDF schemas are commonly used in libraries.
- BIBFRAME
- The Bibliographic Ontology
- DBPedia Ontology
- Dublin Core
- FOAF (Friend of a Friend)
- MADS
- PREMIS
- RDA Unconstrained
- MARC Code List for Relators
- Schema.org
- WikiData
Metadata Mapping Tools
The metadata mapping tool(s) you use will largely depend on your front-end framework, choice of migration tool(s), and requirements. The Islandora community has produced several tools for migrating data and metadata to Islandora 2.0. Each of these tools uses a CSV to map MODS fields to RDF. Please consult the documentation for each tool for in-depth guidance on how to use it.
Metadata Remediation
Metadata cleanup, or remediation, is an optional task that some institutions opt to pursue prior to a migration. In many cases, metadata fields have been used incorrectly or inconsistently over time, and a repository migration is a good opportunity to remediate these problems. However, it is very important to scope these efforts in the context of timelines and available resources; metadata remediation can often take as much time as you give it, and the work is never really done.
...