|
|
Technology, Scalability, and Metadata (or Not): Research Challenges for Digital Libraries
Carl Lagoze, Cornell University
Clifford Lynch, Coalition for Networked Information
William Mischo, University of Illinois at Urbana-Champaign
Moderator Gerry Hanley, NSDL Policy Committee
Notes - Technology, Scalability, and Metadata (or Not): Research Challenges for Digital Libraries
Cliff Lynch
Metadata are assertions made by somebody about something, but not the same as truth
Google has enough trouble with the misrepresentations within the content to worry about verifying the mistruths asserted in metadata.
Do different people following the same rules for cataloging produce the same results?
Affordability
- there is far more content than people to catalog it
- What should be the scale of a learning object? Full course vs. smaller bits increases reusability, but becomes unscalable in terms of cost to catalog.
Suggests integrated application of three technologies
- human description
- content-based retrieval (e.g., speech and image recognition)
- social-based retrieval (e.g., rating systems, personalization, recommender systems)
We need to think about communities that are a community of one.
Bill Mischo Digital Library Issues and Trends
Great distinction between digital libraries and digital collections distinction lies in the integration of digital collections and services via standards and protocols
Great expense dedicated to current scholarly communication systems maintaining libraries and producing prestige journals
We live in a very distributed information environmentmultiple heterogeneous information repositories, portals, commercial providers, hidden web
Carl Lagoze
Paradigm mapping suppliers supply intermediaries who supply consumers
- Faulty Assumptions about this model (slide)
- We need to move toward a more participatory environment greater universal sharing of expertise across roles this changes the paradigm
New assumptions (slide)
- Roles are fluid and ambiguous
- Machines are users too
- Collections boundaries
NSDL Phase I imprinted first set of assumptions and paradigms.
- Questioning educational impact (are mass of resources the answer?)
- How do we determine quality (who determines it? Panel of experts? User?)
- How do we match current norms (what is happening in the learning environments of the users? Chat, blogs, smart mobs)
NSDL as it might be
- We provide mass of information in a way that interrelates the information (data warehouse)
- highly relational environment
Rethinking the role of metadata, not for discovery but more for refinement, when context for resources is provided or for small stand alone objects, is additional or any metadata really necessary? Would full text indexing be better?
Questions
Are there examples of that provide high recall without expensive metadata?
Comments
Please enter any comments in the following format.
- (commenters' initials) - month/day [comment date]
- comment
(pac) - October 14
Can you post Carl Lagoze's notes?
|
|