ER&L 08: A Matter of Semantics?

Mason Hall
Electronic Resources Librarian
Jonathan Blackburn
Web Development Librarian
Florida State University

"A matter of semantics? Intelligence, open data, and the future of ERM"

If we had the perfect database record for an online journal or database, who would maintain it?
Librarians at FSU?
Librarians other places?
Publishers and vendors?
Everyone!

Why hasn't a system like this already been built? The presenters are still exploring this question, and working on how it might look.
The big problem is duplication of data - across knowledgebases, libraries, etc - same information presented many ways.
The fact that each data set has multiple owners might/might not be a problem - individuals don't have to maintain, but are beholden to the data supplier for robustness and accuracy.
Examples of cooperative databases?
Jake - now defunct
CUFTS/GODOT - Simon Frasier U. OS link resolver and KB
LibraryFind - open source link resolver and fed. search
Registry projects like Ockham, IESR, ORCA; possibility of combined global efforts

What would the "theoretical" perfect system look like?  More importantly, what would it be called?
Library OKRA (One KB to Rule them All)
It would be open - as in open access, open source, open data
It would be shared - a lot of different groups can maintain data, so a lot of different groups should maintain
    Models for sharing: Cooperative cataloging; Wikipedia; Freebase.com
It would be semantic - huh?
    Semantic technology is machine-readable; about relationships (like triples: publisher publishes journal; vendor owns database)
    The old way: Item (one) has Attributes (many) - breaks when look for relationships between attributes
    Semantic way forms a web of relationships among many entities; machine can infer these relationships. The idea is that semantic technologies are better than traditional technologies at finding relationships between/among things - finding relationships that the data originators didn't foresee.

A Semantic ERM?
Many, many elements or entities and relationships we track to keep hold of all the information about a journal - a semantic system would track the relationships better than current systems.

Challenges:
Scope - there's a reason we don't already have one of these - it's huge
Privacy - institutional or proprietary information
Metadata schmetadata - hard to imagine how a narrowly defined metadata scheme would work with this project - the metadata allowed would need to be extensible
Incentives? - why would libraries want to do work on this?

The Killer App - what would need to popularize the project/system and gain buy-in:
Is there hope? Yes!
Encourage publishers to be more open with data
Press vendors to uncouple data from services
Support Open Source library software

Continue the discussion on http://LibraryOKRA.com

[Commentary]: One of the best presentations I've seen so far in terms of not showing something we've all seen; bravely speculating about a somewhat pie-in-the-sky project; not sacrificing big ideas on the altar of nailing down every detail. Bravo, gentlemen. Count me in.