Blue Core

A Community-Operated Shared BIBFRAME Datastore

AKA LD4P3 Phase 2

Tom Cramer, Stanford University

Simeon Warner, Cornell University �

12 December 2023

CNI

Conclusion

• BIBFRAME and linked data are the future of bibliographic description
• The LD4P grants & new developments in our systems environment have brought us to an inflection point
• We are here to introduce Blue Core, a project to develop a truly shared bibliographic store of linked data,
• supported by rich tooling and connections across the ecosystem.

History of LD4 Grants

• LD4L (2014-2016) – Exploration, use cases & ontology work
• LD4L Labs (2016-2018) – Tooling for libraries & more ontology
• LD4P (2016-2018) – Sketching metadata workflows & ontology extensions
• LD4P2 (2018-2020) – Sandboxes: cataloging, conversion & authority tooling
• LD4P3 (2020-present) – “Closing the Loop”

Progress Made

• BIBFRAME 2.x
• Sinopia (linked data cataloging tool)
• QA (authority look up & cataloging aid)
• ShareVDE converter & PCC Data Pool
• Discovery (knowledge panels & browse interfaces)
• PCC engagement (Program for Cooperative Cataloging)
• LD4 Organization & annual conference

The world enriched with library data; libraries enriched with the world’s data.

But…

• Why does every institution need to mint its own set of URIs?
• Is it realistic that every institution run its own RDF store?
• Are we mindlessly recreating the MARC environment by copying metadata to edit it locally?
• Isn’t the power of linked data in authoritative, shared URIs with enriching links?
• Is this an opportunity to create a 21st century metadata ecosystem with new models (BIBFRAME), new tooling, new workflows and new efficiencies and affordances for discovery?

​

MARC - A Spectacularly Successful 58 Year Old Standard

​

Development started by Henrietta Avram 1965 at LC, US standard by 1971, MARC21 in 1987, MARCXML in 2002. We all rely on it!

Fireworks by Warren Tobias: https://commons.wikimedia.org/wiki/File:Fireworks_%28Unsplash%29.jpg

BUT… MARC combines in one record information about different entities that should be managed separately.

​

AND… MARC doesn’t use identifiers well for linking or for controlled vocabularies. Too many strings. (Linky-MARC approaches are a stop-gap.)

Rain by W.carter https://commons.wikimedia.org/wiki/File:Rain_splashes_on_rooftop_4.jpg

BIBFRAME for New Growth!

​

1. Properly separates key entities
2. Designed to support migration from and to MARC, necessary for inevitably lengthy transition

Cherry blossoms at Olin Library, Cornell University.

BIBFRAME: Share and Reuse

Linked Data Authorities

Shared Bibliographic Description

Minimal Local Data

What are we building?

• A shared linked datastore for BIBFRAME Works, Instances, and more
• Maintained and operated by a consortium of libraries that is seeded by Cornell, LC, Penn and Stanford, and structured to grow
• That integrates with bibliographic systems and providers
• Like EBSCO, Ex Libris, OCLC, ShareVDE, et al.
• That breaks the “institutional copy” model of cataloging and moves to shared data norms
• And locks bibliographic data open for reuse, accessible to other institutions

Why are we building this?

• Avoid duplication of effort
• Allows linked data to work at scale
• Allows libraries to better focus their efforts towards making richer knowledge linkages
• Enables data use by and (longer term) contributions from small and diverse libraries
• Lays the foundation for a new future where a diverse set of institutions work together to incorporate and apply linked open data cataloging
• Avoids vendor lock-in associated with data, systems, �or services

Blue Core

Architecture

Traditional workflows in and out, different store

Local discovery at each institution driven from shared store data

Entity focused RDF

cataloging

Additional Considerations

1. Workflow & Dataflow
1. What do cataloging workflows look like with Blue Core?
2. Integrations with ILSes, editors, et al.
3. Dataflow, system of record, updates across the ecosystem
2. Organizational Design
• How to define cataloging norms for a truly shared pool
• Governance
• Resourcing for build out & sustainability phases
3. Project Details
• Participation from catalogers, libraries, commercial sector, et al.
• Project structure, plan and timeline

Approach

• Iterative
• Success in linked data initiatives often turns out different than originally envisioned
• Developing an ecosystem, not a single system
• Build in time to assess and adapt to findings
• Incrementalist
• Every phase a baby step
• Each with its own value
• Minimize risk
• Prototype and build the tooling as we go
• Cataloging can’t progress if work is only conceptual

Timeline

2023 – Conceptualize

2024 – Plan & prototype

2025 – Move to implementation

Discussion