MSC feedback on DPLA metadata application profile update (4.1)
Summary of MSC feedback to DPLA
The Metadata Standards Committee (MSC) sees that the standard is not a standard for creating new metadata but a reformatting of the metadata that DPLA is harvesting. As this metadata is not intended for human creation but rather automatic mapping we believe that the appropriate level of feedback is on the specific changes proposed in the 4.1 revision. Overall, the application profile approach of DPLA aligns well with the principles that the MSC promotes. Given this close alignment the feedback below does not dive into detail where there is good alignment.
The principles highlighted by the MSC focus on higher level issues related to metadata including design and use implications. As such the feedback provided here does not dive into the technical details and changes in the MAP 4.1 revision.
Overall, the recommendations below affirm the direction of the 4.1 MAP and the overall approach of DPLA in developing metadata standards.
1. Metadata standards should be part of a shared data network
Principle context: Metadata—its standards, systems, and services—is most efficient when connected with the broader network of information as manifested in the W3C’s Linked Data initiatives and related data sharing efforts. Metadata standards should provide mechanisms for expressing relationships to other vocabularies. For example, standards that allow for the use of URIs, in addition to or instead of relying solely upon strings of text, may connect information from different sources, support distribution to indexing and research services, and increase resource visibility on the web.
Feedback: The DPLA, almost by definition is a shared data network and this application profile meets this principle. No specific feedback provided here is specifically related to the 4.1 update. The fact that DPLA relies on existing resources (e.g. the EDM, MODS and other schemas and vocabularies) is viewed as a positive trend.
2. Metadata standards should be open and reusable
Principle context: Open metadata is a foundational building block of information systems and computationally-enhanced research. Metadata standards and associated vocabularies need to be open for use and re-use, free of charge, and managed with openness and transparency. Metadata standards governance and maintenance bodies should consider explicit licenses that promote reusability of standards and vocabularies.
Feedback: The DPLA’s use of namespaced vocabularies promote reusability and computational analysis. The fact that the MAP is published along with supporting documents with alternative descriptive information are viewed as a good mechanism to promote this make it easier to re-use. We note that in some cases, the MAP collects data as literal values without a specific definition of the source vocabulary (e.g. the hierarchicalGeographic). We wondered if there were cases where content providers would provide URIs rather than literal values and if this would influence the behavior of the schema (e.g. would URIs be de-referenced before inclusion). Likewise, although not necessary, we wondered if some statement around the ability to re-use the metadata standard itself would be helpful.
3. Metadata standards and creation guidelines should benefit user communities
Principle context: The volume of information in a web-enabled world, along with the ability of information systems to analyze and index digital objects themselves, changes the value of traditional metadata. Metadata standards should be created with the intended audience, its related communities, and the public in mind. The depth and granularity defined by metadata standards should prioritize data elements that support clear use cases defined and contributed by their user communities and allow implementers to use only features relevant to their needs.
Feedback: The focus of the MAP on consolidating multiple metadata schemas into a single shared platform is perceived as a positive step towards sharing and community use. By using standards and platforms that LAM communities are familiar with (e.g. DC, DCTERMS, MODS, EDM) we believe that the MAP is positioned to be more successful than it would with an independent schema or vocabulary. Specifically separating licenses from rights statements is viewed as a positive development in the MAP. We are unsure if there are types of objects that are more likely to have licensing rather than rights however and collectively had difficulty identifying clear use cases for this more granular metadata. As such more definition on the intended use or role of the license and rights fields could be useful. The group debated for example if there were cases where a rights and license statement would be used together rather than in some hierarchical fashion. Perhaps one argument against this approach is that having separate fields in this case could introduce complications around derivative uses.
4 . Metadata standards should support creative applications
Principle context: Up to the 21st Century, bibliographic metadata has supported a relatively narrow vein of research, involving reading, viewing, listening to, or otherwise engaging with individual resources by humans. As new research methods emerge (e.g., computational linguistics, computational bibliometrics, linguistic analysis, network analysis), and as new data modeling methods geared toward machine actionability mature, metadata standards and exchange/access methods should easily support evolving modeling and use practices. Encoding of metadata standards in machine-processable formats such as RelaxNG, XML Schema, or RDF Schema can help to promote computational use of metadata in emerging fields.
Feedback: The 4.1 MAP version appears to extend and improve creative application and extended uses. We note that the standard uses RDF but could not readily find information on how the standard is made available (e.g. we could only find the PDF and excel version). A published schema in RDF, XML or RelaxNG might be helpful. Likewise ensuring that there is a URI that points back to the full metadata is helpful for creative applications.
5. Metadata standards should have an active maintenance and governance community
Principle context: Metadata standards are only as valuable and current as their communities of practice, which are changing with increasing rapidity. The needs of implementers should be a driving force for the development of standards over time. Governance bodies should support transparent, timely development and revision processes by publishing standards development and revision protocols. Such protocols should include mechanisms for community contributions to draft standards. Governance bodies should change and develop over time to reflect the diversity of their current and potential user communities.
Feedback: The DPLA is not only active but appears to maintain ties to other active communities. The MSC notes that the call for review was widely distributed and that the standard itself had received feedback in the form of comments. It would appear from these activities that DPLA is not just active but actively contributing to the standards.
6. Metadata standards should be extensible, embeddable, and interoperable
Principle context: Central to the successful deployment of a metadata standard is its ability to be used effectively alongside other standards. Standards should be designed in a modular way to allow relevant parts to be incorporated into local systems together with parts of standards from other sources, with each given equal preference. Relatively complex standards should provide “lite” style implementation options, to allow for wider adoption and utility, even if this means lossy data transfer in some cases. “Lite” style implementations are helpful in enhancing interoperability of metadata standards that have differing underlying intellectual models.
Feedback: We note that the MAP itself is a “lite” version of the metadata that DPLA might have to work with and, from the committee’s read, appeared to strike a good balance between complexity and adaptability. The fact that a large amount of the terms map to some external vocabulary is viewed as positive and the reliance on optional metadata we believe is a good mechanism to spur adoption and participation. It could be that establishing a rights vocabulary (if possible) would enable higher scale derivative uses or simplify computational decision making around derivative use. One potential method would be to list potential vocabularies to encourage adoption (e.g. rightsstatement.org)
7. Metadata standards should follow the rules of “graceful degradation” and “progressive enhancement”
Principle context: Metadata standards should be designed such that, as the complexity of the standard is stripped away, the metadata degrades gracefully. Removing complexity, for example to make metadata that is encoded according to a robust standard useful to a less specialized audience or to remove sensitive information, should result in metadata that is still useful. Metadata standards that are designed from a progressive enhancement approach prioritize simplicity while being flexible enough to support enrichment of the standard to accommodate specialized metadata use cases.
Feedback: The MAP appears to follow this guideline.
8. Metadata standards should be documented
Principle context: Alongside the metadata standard, information about its use in practice, examples of the standard as applied to a real-life resource, its governance and management structure, rationale behind design decisions, and its connections to metadata standards from other domains should be published. Efforts should be made to allow easy adaptation of documentation for multilingual environments and to make the documentation understandable to implementers from other fields. Applying a license to the documentation will clarify its use and reuse.
Feedback: We believe that it would be useful to make recommendations for vocabularies, even if they are not required (e.g. we note that AAT is cited as one vocabulary but many other elements would benefit from vocabulary recommendations as well). This might create an opportunity to support LAMS and encourage overall higher quality metadata by including these recommendations, even if not as a specific part of the MAP.
9. Metadata standards should be inclusive and transparent about historical and cultural biases
Principle context: Metadata standards development is not neutral; human beings unavoidably assign value judgments when making (and not making) assertions about a resource, and in defining the assertions that can be made about a resource. Metadata standards developers should be aware of these value judgments, make them explicit to the degree possible, and take as a guiding principle not neutrality, but rather inclusivity of worldviews. A diversified team approach can be considered in the creation, implementation, and further enhancement of the metadata standards. Metadata standards and vocabularies should reflect changes in language.
Feedback: We note that as a consolidated application profile the risk of loss of context is a key issue around inclusivity and transparency. Providing access back to the object and source metadata is a good mechanism to mitigate this risk. We also note that a lack of vocabulary recommendations or requirements is useful in encouraging and supporting a wide range of adoption and actually may help address this challenge. Finally, by having so few requirements, this MAP opens the number of communities who can participate in the DPLA. We believe that this could further encourage participation in metadata aggregation and collaboration across libraries, especially for LAM institutions of multiple sizes and resource scales. Given that the focus of the DPLA is America, some Western-centric orientation in the geographic portions is sensible, but the committee wondered if there were alternative viewpoints (Native American or Inuit, for example) that have been overlooked in forming the structure of the metadata schema or vocabularies.