Lorcan Dempsey of the OCLC has a fascinating article on what the long tail means for libraries. For example, 10% of titles in libraries account for 90% of loans. But, users of services such as Netflix go deep into the catalog. What can libraries do, Lorcan wonders, to help patrons find what they don’t know is there and they may not know they’re interested in? (I’ve put this too simply…) [Tags: libraries lorcan_dempsey long_tail oclc ranganathan everything_is_miscellaneous]

3 Comments »

D-Lib Magazine has a special issue on digital libraries. I’ve just started reading it, but so far it’s fascinating. (Thanks to Kurt Starsinic for pointing it out.) [Tags: taxonomy everything_is_miscellaneous EverythingIsMiscellaneous libraries]

Be the first to comment »

Jakob Voss has published a paper on Wikipedia’s category system and how it compares with tag-based folksonomies and the Dewey Decimal Classification system. At the time he did the research (Jan., 2006), there were 923,196 articles in the Gnlish version, and a full 94% of them had at least one category tag. There were a 91,502 unique categories. Then the paper gets statistical beyond my understanding, but that’s my problem, not the paper’s… [Tags: wikipedia tagging taxonomy everything_is_miscellaneous]

2 Comments »

In a sidebar to an article on the problems the pure physicality of books causes people with lots of them (er, books, not problems), Marie Arana and Jonathan Yardley write about their family argument over how to sort books. She is the editor of Bookworld. He is the Washington Post’s book critic. She sorts them in an arbitrary pattern “that makes perfect sense to her but to anyone else is an unfathomably mystery.” He keeps nonfiction upstairs and fiction downstairs; he sub-sorts by category and arranges biographies alphabetically by subject.

If you have an interesting, ingenious or particularly dumb way of sorting books, I’d like to hear about it. (I’ll assume you’re ok with my using it as an example in my book or somewhere else.) [Tags: taxonomy EverythingIsMiscellaneous everything_is_miscellaneous marie_arana jonathan_yardley books]

19 Comments »

Here are some gorgeous old card catalog cards. (Thanks to Ben Hyde for the link.) [Tags: everything_is_miscellaneous card_catalog libraries ]

Be the first to comment »

Donna Maurer of Maadmob Interaction Design has a session explaining to IA’s why they ought to take seriously George Lakoff’s Women, Fire and Dangerous Things, a book basically about Eleanor Rosch’s prototype theory. (I’m about to start writing about this in my book.) It’s a difficult book to read, despite its great title.

Categories, she says, have been taken as being existing/natural divisions that can be expressed in clear definitions. But, Lakoff says (says Donna), categories have prototypical examples. We can organize by similarities to the prototypes without having to define the category.

Donna talks about the concept of basic level categories— you generalize up from them and down to the specifics. BLCs are the level you learn first and usually has a short name, e.g., “dog.” It is the “highest level at which a single mental image can reflect the category.” [In English we can’t form an image of furniture but we can of chair.] BLCs depend on the culture. E.g., city dwellers will say “a tree” while a country dweller might say “a maple tree.”

Donna looked for a web site that organized information by laying out BLCs, but she couldn’t find one. E.g., the eBay site’s category list mixes BLCs with more specific categories.

Prototype theory is important to IAs because the classical theory of categories is built into much software. She says this is her most important point: “Recognize that categories occur and you’ll be less stressed about categorization that is not neat.” Messiness is ok. No scheme is going to include everything and be perfect. (Someone from the audience points out that Sotheby’s “Other” category has an “Other” sub-category.)

Other implications for IAs. Donna suggests using less prototypical items to describe edge cases. And, she says, you can derive BLC names by doing research to see what they are for your users. “Basic level items are easily recognized and likely to have good scent.” Card sort, she suggests, with basic level items rather than more granular content elements. “Get people to the basic level of the hierarchy as soon as possible.”

She wonders whether tags are often BLCs, accounting for their popularity. (Livia from the audience says that they probably are basic level if the taggers are typical of the group.)

Q: Does Lakoff deal with the cultural differences?
A: Definitely. That’s what the title is about.

Q: Folksonomies can help us get away from us imposing a taxonomy on a system.

Q: The same category may be basic level for one group and high level for another…

Q: (Christian Crumlish) When you focus on basic level categories for navigation, where should they be on the site, typically?
A: On the home page, if possible.

Q: (me) How stable are BLCs and what size group do they vary over?
A: They’re pretty stable over time, and the size of the group depens on the domain.

[Excellent talk. Exciting to me to see IA’s taking this stuff up.] [Tags: iasummit donna_urer george_lakoff prototype taxonomy everything_is_miscellaneous eleanor_rosch]

4 Comments »

Continuing what is apparently my new policy of only blogging about Wikipedia, with occasional posts about how Bush sucks: Jay Fienberg recommends that we take a look at the Semantic MediaWiki if we want to see the Semantic Web in action. He says that it even shows that RDF can be easy to use.

He also points out (in an email) that Pandora, a music taste site that’s got lots of people excited, uses a “ginormous faceted taxonomy/ontology.” (Just to make my position clear, if I actually have one: Local ontologies obviously can be highly useful. The question is how hard is it to knit them together reasonably well? And then the question is what counts as reasonably well. But one thing we know for sure: Bush sucks.)

Finally, Jay recommends a post by Steve Krause about Pandora vs. Last.fm. “Pandora’s recommendations are based on the inherent qualities of the music,” says Steve, while Last.fm is a collaborative filter. [Tags: jay_fienberg steve_krause pandora taxonomy ontology everything_is_miscellaneous EverythingIsMiscellaneous semantic_web wiki]

And while talking about taxonomies, VerbDate.com is a new, free dating site that integrates tags, tagclouds, Skype and google maps. If it were any more Web 2.0-ish, you’d only be able to date Tim O’Reilly.

2 Comments »

Jeneane posts about using tags to provide commentary on and context for a post, the way some magazines and blogs use headlines to comment on the text. (Doc Searls sometimes does this, and Esquire’s Dubious Achievements used this effect to great comic effect.) Such tags may not help us find the posts, but they can help us understand them. They can also make us laugh. [Tags: taxonomy EverythingIsMiscellaneous jeneane_sessum tagging]

3 Comments »

Tara Hunt is compiling reasons why people tag. To her list I’d add: So readers can read only the topics on my blog that they care about (which theoretically should lower my inhibitions about blogging “off topic”) and to increase (a tiny bit) the world’s meaning and/or intelligibility.

Go add your own reasons…

Be the first to comment »

Emanuele Quintarelli suggests that we go further than Peter Van Dijck‘s mefeedia that provides a few broad facets by which tags can be organized. Why not create a wiki at which users can create a much more detailed, hierarchical taxonomy that would know, for example, that Venice is not just a place but that it’s a place in Italy (as well as a place in the US)?

It might work, but it raises three key questions when it comes to tags and taxonomies:

1. Tags have succeeded because they require so little thought and can used without regard for their meaning beyond oneself. (I suspect many of us do consider social meanings, but one doesn’t have to.) Would a wiki that requires us to think about metadata work? Would enough people participate?

2. If enough people participate, could we come up with a universal taxonomy specific enough to be useful? Do we think about things in sufficiently the same ways?

3. If such a taxonomy existed, would we use it? It would be a bottom-up controlled vocabulary, but it’d still be a controlled vocabulary that requires people to look things up.

I suspect that having reduced the problem of metadata to its most elemental form — type in a word or two that will remind us of what a page or photo is about — we will now complexify it usefully to the point at which the complexity gets in the way. It’s impossible to predict where those various points will be. (Complexity reaches its own level of misunderstanding.) I thus don’t think we can predict where Emanuele’s suggestion falls, although I worry that it’s more appealing to information architects than to “normal” people. (Yeah yeah, info architects are normal people. Except not when it comes to information. Besides, have you ever hung out with information architects??? :) [Tags: everything-is_miscellaneous EverythingIsMiscellaneous information_architecture ia folksonomies taxonomy]

30 Comments »