Slideshow Transcript

1. Slide 1: Digital Library Infrastructure for a Million Books: Synthesis (and Opinion) Steve Toub California Digital Library
2. Slide 2: Having books talk to each other in the background • Ability to enable books to talk to each other – Too costly to pre-define all relationships a priori – Need a system that grows, learns dynamically • Requires: – Rich markup of high-value texts – Ability to expose structure (e.g., microformats) OR address latent structure • Including well-managed identifiers within a text – Ability to define relationships between hooks in one text and one in another • If not RDF/OWL, something conceptually similar – Tools, APIs (standards,service registries, later)
3. Slide 3: “How many libraries do we need?” • We must operate under the assumption that that everything will live at the network level • Dividing the heterogeneous million book corpus in to sub-corpora, by domain, is necessary to apply tools effectively • Unless things change radically, humanities scholars, libraries, and the government don’t have the resources or will to shape how digital library services will exist in an open, interoperable way at the network level
4. Slide 4: “Explore effectiveness of discovery tools” • Listed only as a single bullet point on one of Bruce Robertson’s slides but something near and dear to my heart • Example: – George Bancroft's History of the United States: Vol. 1, History of the Colonization of the United States
5. Slide 5: Bancroft’s History of the US: Perseus
6. Slide 6: Bancroft’s History of the US: MoA
7. Slide 7: Bancroft’s History of the US: MoA 2
8. Slide 8: Bancroft’s History of the US: Google
9. Slide 9: Bancroft’s History of the US: Google 2
10. Slide 10: Bancroft’s History of the US: Google 3
11. Slide 11: Bancroft’s History of the US: Google 4
12. Slide 12: Bancroft’s History of the US: Google 5
13. Slide 13: Bancroft’s History of the US: IA
14. Slide 14: Bancroft’s History of the US: IA 2
15. Slide 15: Bancroft’s History of the US: IA 3
16. Slide 16: Bancroft’s History of the US: IA 4
17. Slide 17: Bancroft’s History of the US: Gale
18. Slide 18: Bancroft’s History of the US: WorldCat Editions
19. Slide 19: Bancroft’s History of the US: WorldCat Results
20. Slide 20: Discovery infrastructure at the network level • Evangelize metadata/content exposure APIs – OAI-PMH, COinS, hCite,RDFa, unAPI, OAI-ORE • Ability to aggregate raw data and scrubbed data at the network level (plus scrubbing tools and loaders) • An authoritative content registry (i.e., OCLC Registry of Digital Works) – Work-level (and expression-level) identifiers – Low-barrier workflows for update + expose/harvest • Formalized ways to express and expose content relationships: xISBN + thingISBN + ML approaches – Related objects (versions, editions, translations, dupes, …) – Compound objects (chapters, overlay journals, …) – References (annotations, quotations, excerpts, …)
21. Slide 21: Need more focus on usage, research process as a whole • Tools like Piggy Bank, Zotero can capture • Count transactions (views, downloads, annotations, citations…) in standardized ways • Standardize log data (e.g., COUNTER) and its exchange (e.g., SUSHI) • Consolidate transaction data to close feedback loops and improve services – Metrics (e.g., MESUR, search analytics) – Ranking, recommending, social discovery – Matching user terms to: authority files; texts
22. Slide 22: Infrastructure for identity/users • OpenID is only a hint of what's to come • Identity is the foundation for higher-level services: trust, authority, verified claims, reputation, groups/memberships, … • How can this emerging ecosystem leverage existing infrastructure? – Authority files – LDAP servers – Peer-review, citations, etc.
23. Slide 23: Can’t take these for granted • Preservation – Anyone who has looked at credible cost breakdown for a well-managed long-term digital preservation repository will understand that the libraries will always preserve NOT a given • Intellectual property rights – GBS not showing many public domain items – Keeping tabs on orphan works
24. Slide 24: Sustainable business models • Crucial to think about economic sustainability and shape realistic economic incentives • Why don’t libraries license JSTOR indefinitely so it could be opened up free for all? • How to incentivize disciplinary collaboration that go across institutions? • What’s our relationship to players at network level: Google, MSFT, IA, OCLC, Amazon, … • Who pays for the common infrastructure? • Organizational infrastructure to: • Enable free riders? Feed and care for free kittens?
25. Slide 25: Where to begin: Any consensus? • Keep increasing available texts? • Compelling applications? • Easy to use tools? • APIs, service registries? • Key software engineering tasks? • Marketing and outreach?