Comments on: metadata

By: Natalia Cecire

Natalia Cecire — Mon, 16 Nov 2009 18:06:26 +0000

Oh, yes, Google Scholar. Enter “Wordsworth” and you’ll get obscure scientific studies with one co-author with the last name “Wordsworth.” There has to be a better way. It’s not just a matter of what they index, but also how. You’re so right that “it’s just not the metadata that we might be most interested in, or that might produce the best results.” There are ways to use metadata that are, for some purposes, close to meaningless.

By: Dorothea Salo

Dorothea Salo — Tue, 06 Oct 2009 20:25:47 +0000

They’re not even making full use of the MARC records that participating libraries are giving them.

By: Kathleen Fitzpatrick

Kathleen Fitzpatrick — Sat, 26 Sep 2009 21:42:38 +0000

That’s helpful, Kari — I meant to bring COinS into this chapter, and I’m not sure why it didn’t finally appear here!

By: kkraus

kkraus — Sat, 26 Sep 2009 20:27:26 +0000

Just a quick addendum: the COinS metadata that Zotero interprets is relevant to your discussion of OpenURL in the next section.

By: kkraus

kkraus — Sat, 26 Sep 2009 19:52:58 +0000

Users are also encouraged to make their personal websites and web resources Zotero-readable by, for example, adding Dublin Core or COinS metadata to their HTML; or by downloading and installing a plug-in for a supported content management system, such as WordPress. See the documentation on “Making Your Site Zotero Friendly”: http://www.zotero.org/support/make_your_site_zotero_ready

By: Kathleen Fitzpatrick

Kathleen Fitzpatrick — Fri, 25 Sep 2009 05:09:21 +0000

Absolutely. I’ve read about projects that are attempting to produce machine disambiguation, but I suspect they’re a fair bit down the road as yet, and will likely still need expert supervision and training…

By: Kathleen Fitzpatrick

Kathleen Fitzpatrick — Fri, 25 Sep 2009 05:07:52 +0000

That’s a helpful clarification, Amanda; thanks.

By: amandafrench

amandafrench — Wed, 23 Sep 2009 16:11:10 +0000

A small additional point about this paragraph: a key benefit of expert-produced ontologies is what librarians call “authority control,” an apparently tautological and tyrannical term that just means one can distinguish Jones, John M. 1911-1987 from Jones, John H. 1987-. Technology can sometimes do this kind of disambiguation, as well, but I can’t think of any really good machine-generated examples, and of course users are likely simply to tag something “John Jones.”

By: amandafrench

amandafrench — Wed, 23 Sep 2009 15:58:41 +0000

Actually I don't think it's true that URLs to individual tweets cease to work, or at least, they haven't yet for some of the oldest tweets of mine that I can get my hands on (about 2 years old, dating from late 2007). The problem is that you can't *find* those tweets or their URLs after a period that is currently only 1.5 weeks. That very short time period is "dynamic and subject to shrink as the number of tweets per day continues to grow"; it used to be about four months. Therefore, the only methods of preserving Twitter are currently very ad hoc, and they have to be set up beforehand (TwapperKeeper, RSS, and so on). Still, if you save the URL of an individual tweet, you can link to it with reasonable confidence.

By: Kathleen Fitzpatrick

Kathleen Fitzpatrick — Sun, 20 Sep 2009 20:11:47 +0000

I was there for that moment, and would also be curious to know why they’re not making use of crowdsourcing for metadata correction. Or even just allowing known librarians to correct the data. I’ll hope that they’re developing a system, too.

In the meantime, yeah, I’m going to have to draw the line somehow on updates, but this one seems pretty key.

By: amandafrench

amandafrench — Sun, 20 Sep 2009 07:50:45 +0000

I think you’re probably right about the need to update this to take Nunberg into account — though I hope you’ve got a strategy for preventing yourself from updating *everything* as new news breaks. And were you there for the moment at DH 2009 when someone in the audience urgently asked Jon Orwant for that very functionality? I twittered it.

I’d be very grateful to learn why Google Books doesn’t allow metadata correction. I suspect they’re secretly developing a system to allow it, frankly.

By: Kathleen Fitzpatrick

Kathleen Fitzpatrick — Mon, 31 Aug 2009 14:51:13 +0000

I think this paragraph really needs to be updated to think about what Geoff Nunberg has called the "metadata train wreck" of Google Books; relying on machine-generated rather than expert-produced metadata will be disastrous -- but even if one starts with that base layer of machine-generated error, it could be corrected via crowd-sourcing.