|deborah (deborah) wrote,|
@ 2007-06-27 10:44 am UTC
|Entry tags:||conferences: jcdl, education, technology, user interfaces|
Much of this panel was dedicated to speakers from the IMLS and the NIH talking about their grant funding projects. A valuable talk, but not something that led me to take many notes. The key part of this panel, for me, was Greg Crane's speech, although I've heard him give versions of this talk before. While I've often disagreed with various points Crane has made, the overarching thrust of this particular talk is what I find very valuable. His focus is not a library focus about digitizing collections for preservation or access, but it's specifically a researcher focus: what tools can we add to our resources as we digitize them to give us more than we ever had before.
Humanities Cyberinfrastructure (Crane)
The grand humanities questions:
How do we understand human expression? And how do we communicate that understanding?
- National institutions and topics, with American and English language dominating
- Global language and culture are crucial
What do humanities scholars want to do?
- Produce more knowledge
- For professional advancement
- But only a few people will understand you
- Enhancing intellectual access
- Humanities scholarhsip isn't accessible outside of the academy, and that's a problem
- We need to open up our ideas to new audiences
- We need to be cross-disciplinary, cross-language, cross-cultural
What might a truly digital environment/document look like? It shouldn't be incunabula, (eg. PDFs, moveable type that looks like manuscript scribing).
- Separation of data from presentation
- Recombinant data
- Books need to talk to one another
What does it mean to usefully create a "million book online library"? Three core technologies are needed.
- analog to text
- language translation
- text to data
He then gave several slides which amount to a logical statement of "let's FRBRize and TEI markup all our books, down to character and quotation recognition and markup, and we need services to do this automatically." Basically, name every entity you can.
Humanists have a bad habit of making crappy, too-specific, unmaintainable software, and we need to fix this.
- Minimal domain-specific technology
- Maximal non-incunabular documents (which will be training sets for other people's systems, and will provoke other software providers to do clever things with your documents)
- Human capital
Ray @ IMLS on the humanities information landscape
Ray worries that many humanities resources (say, in history) are analog. The money to digitize is hard to come by. The assets do not lose value over time and they must be preserved, but by whom? And how can we guarantee the public's right to know? How can we create trusted repositories with authentic records?
Collaboration is hard, but we must do it.
Humanities projects should publish "lessons learned" white papers after both successful and failed grants are complete. The sciences accept that grants might fail; the humanities don't, so they hide the failed projects, and nobody gets to learn from the experience.