What systems for organizing knowledge exist?

February 20, 2006 12:22 PM

What systems for organizing knowledge (other than the Dewey Decimal and Library of Congress) exist?

I'm particularly interested in anything obscure or hare-brained.

posted by richard m to writing & language (20 comments total)

I've been fascinated by the concept of the memory palace ever since I read about it in Hannibal.
posted by Gator at 12:25 PM on February 20, 2006

There are many, and I expect the MeFi Librarian Squad will be along shortly to elaborate. The Wikipedia library classification is a good place to start. I've always been interested by Ranganathan's colon classification.

Here is a lengthy, though not exhaustive, list of both classification systems and controlled vocabularies (i.e. subject heading lists, like the Library of Congress Subject Headings, which are similar to but different from classification systems).
posted by IshmaelGraves at 12:30 PM on February 20, 2006

Might not be exactly what you're thinking of, but don't forget folksonomy.
posted by electric_counterpoint at 12:47 PM on February 20, 2006

There's this fictional one, invented by Borges in one of his essays. Borges was the director of the Argentine National Library, and shared your fascination with anything obscure and hare-brained, so he knew what he was making fun of.
posted by Aaorn at 12:56 PM on February 20, 2006

Don't forget about metadata like tags!
posted by parallax7d at 12:58 PM on February 20, 2006

You might like this previous question about ways to organize a personal library.
posted by teleskiving at 1:00 PM on February 20, 2006

The orginal Roget’s Thesaurus had an interesting way of classifing words/ideas.
posted by bigmusic at 1:04 PM on February 20, 2006

The two I remember briefly touching on in library school are Ranganathan's colon classification (also note the other classification links on that page, which provide additional information) and the Cutter's Expansive Classification. (I worked at a library at a major research university in the 90s. Although most of the collection was LCC, there was a (relatively) small, old, rarely used set of books classified using Cutter, which no one had ever bothered to re-catalog.)

"In Book II of De Dignitate (his expanded version of the Advancement) [Francis] Bacon outlines his scheme for a new division of human knowledge into three primary categories: History, Poesy, and Philosophy (which he associates respectively with the three fundamental 'faculties' of mind – memory, imagination, and reason)." Thomas Jefferson organized his library around this scheme.

IshmaelGraves's second link includes many examples of subject-specific classification. I don't know if you're only looking for systems which try to classify all human knowledge, or if you're interested in the narrower ones too, but if you are, one of the more esoteric ones has to be the Encyclopedia of Chess Openings (ECO) codes, dividing chess openings into 500 groups (A00-E99) and widely used in chess literature. (See just the specification or a more detailed list which correlates the codes with opening names and common variations within each code.)
posted by DevilsAdvocate at 1:13 PM on February 20, 2006

The Yahoo classification maybe? The one by Borges is my favourite.
posted by rongorongo at 1:17 PM on February 20, 2006

Elaine Svenonius wrote The Intellectual Foundation of Information Organization, which is an excellent, if rather dry, overview of classification theory. Reading this book will give you the tools to understand most classification schemes. To understand the rest of them, I'd look to Women, Fire, and Dangerous Things, by linguist George Lakoff. His approach looks more at how we mentally create categories and what determines the likeness of like things.

I'd take a look at this Ask.Me on taxonomies, thesaurii, and ontologies as well.

IshmaelGraves excellent link gives you an idea of just how many classification schemes are out there. There are a lot more that aren't exposed to the user. Corbis, for example, has an absolutely bizarre thesaurus that is intended to allow the user to search on the emotional content of an image. The idea is that if an ad agency art director says, "get me a photo that's edgy," the production term can search on edgy and find what they're looking for. Couple that with the fact that they support searching in 4-5 languages and you get a classification system that is pretty dang complicated.
posted by stet at 1:24 PM on February 20, 2006

Incidentally, Umberto Eco's [nonfiction] The Search for the Perfect Language spends a lot of time discussing various attempts from the Middle Ages and the Renaissance to develop languages based around taxonomic classification — so the first syllable of a noun might establish it as referring to something animate, the second as being a mammal, the third as having four legs, etc. None of these worked very well in practice, but it's fascinating reading for someone interested in hare-brained schemes of knowledge classification.
posted by IshmaelGraves at 2:22 PM on February 20, 2006

There's lots more on the wikipedia page on Knowledge Representation.

Notably absent from above discussions so far are mind maps, which are an unusually unconstrained system, typically used by individuals or small teams to allow the visualisation of relationships between concepts around a specific subject. A major advantage of this system is you get you draw pictures and use different coloured pens.

The dewey and colon systems are faceted classifications, but the idea can be (and is) implemented far more dramatically in digital systems, where the items stored are virtual and the limits are massive. A typical implementation is Amazon, which uses many facets to enable users to look for items in different ways, for example genre, author, similar authors, popularity etc. More about faceted browsing.

More esoterically, some consider the I Ching to be a representation of all knowledge, or states of being (or change). I don't pretend to understand this concept, but it is intriguing.

Gnod is an interesting experiment in relating books and music by aggregating individual people's tastes. Check out their music map for the Beatles.

But if you're looking for hare-brained, I am experimenting with an uncontrolled heterarchy of links, the idea being that the system is intended to change over time. I started making it because I have several thousands of bookmarks on many overlapping subjects, and this seemed like the only way to store them usefully (I don't enjoy tagging). There isn't much to look at now, but there is a link in my profile to see a simple heterarchical directory of links about the game of Go.
posted by MetaMonkey at 2:33 PM on February 20, 2006

There's also the tried-and-true "sorted by piles" methodology. It's the system I was using until a few weeks ago.

Now I'm using the "move it closer" methodology, in which I move things closer to where they will be (or, when they're close enough, actually put them away.)

Soon I hope to use the "it's easier to put it away than leave it out" methodology. I anticipate great things.
posted by five fresh fish at 2:50 PM on February 20, 2006

In no particular order (ha):

1. The human brain, which organizes knowledge in such a way that not even the human brain can figure out how it is done. Also, the retrieval mechanism is wonky and unpredictable (see Proust).
   
2. The universe. Everything seems to run in an orderly and unambiguous way, yet no single classification scheme describes it satisfactorily. It is a maddening philosophical quandary that only librarians have the strength of will to look at squarely in the eyes on a daily basis. My God, what giants.
   
   • May be related to #1. Maybe it's just the way our brains work that makes us think the universe has a natural order that we can reverse engineer in the first place?
     
3. Grocery stores: How are they organized? It's a mish-mash:
   
   • By food group
     
   • By meal
     
   • By activities (baking, cleaning)
     
   • By temperature (ice cream is next to frozen pizza)
     
   • By ergonomics (heavy stuff on lower shelves)
     
   • By store layout (the packaged meats need to be along a wall because the butcher is close by, and he needs room to work)
     
   • By popularity (most popular products in most prominent locations)
     
   • By age group (c.f. the cereal aisle)
     
   • By company (Lay's potato chips are grouped, so are Quaker Oats, so are Kraft cheeses)
     
   • By foot traffic (what path do people naturally take through the store?)
     
   • But in the end, what it's really organized by is simply by what works to sell the most product. If moving something to the other side of the store correllates to better sales, a store will keep it. So, there's an certain evolutionary process going on as well.
     
4. Personal book shelves. Whereas library book shelves are organized linearly (Dewey is a line extending from 000 to 999), personal book shelves are often organized basically in a radial pattern. That is, as you take your books off the shelves and put them back in after reading them, the books you access most often are in the center, because that is where you're most likely to shelve them if you're in a hurry, or just don't care.
   
   As you move outward from that center point, you find books that don't get moved as often. Compare distance from the most convenient location on the bookshelf to the frequency of use and you'd probably get a normal curve, assuming the shelver is sufficiently lazy.
   
   Or maybe not. Just a pet theory of mine that I like to make people indulge.

You're probably looking for more traditional classification schemes, though, so you might check out the Universal Decimal Classification, which is like Dewey, but with some interesting notation that lets you talk about relationships. I don't see that anybody has mentioned that one yet, and it's a worthy effort.
posted by Hildago at 5:38 PM on February 20, 2006

Richard Saul Wurman's book Information Anxiety and its sequel have great discussions about how we process and organize information. One of the pieces I found most interesting was the discussion related to sorting, finding the way in which new data connects to what you already know.
posted by sciencejock at 8:00 PM on February 20, 2006

John Wilkins’s Essay Toward a Real Character and a Philosophical Language, which he compiled in the 1660s, was a concerted and thorough attempt at compiling a universal taxonomy, with a view to creating a universal language. It’s one of the classification schemes discussed in Eco’s In Search of the Perfect Language, that IshmaelGraves mentioned above, and is the subject of the essay by Borges in which he brings ‘a certain Chinese encyclopaedia’ to our attention, as mentioned here by Aaorn and rongorongo. Wilkins’s essay, moreover, apparently provided Peter Mark Roget with a good deal of the classificatory framework in his original Thesaurus, which bigmusic noted above. The full text of the Essay is on-line here.
posted by misteraitch at 12:06 AM on February 21, 2006

I just read an interesting article (from a biology point of view, but still relevant) that may be of interest -
Ontology is Overrated: Categories, Links, and Tags - that made some interesting points on methods of classification.
posted by primer_dimer at 2:36 AM on February 21, 2006

Clay Shirky certainly fulfills the "hare-brained" criterion.
posted by stet at 4:28 PM on February 21, 2006

Oh snap no you din't!

But yeah, Ontology is Overrated is itself pretty overrated, in my opinion. The idea that folksonomies were going to save us all was a little silly.

What we've learned is that tags are good for the things they were implemented to do: help you find your own stuff again at a later date, and browse for interesting content tagged by likeminded people. Useful, but not the panacea Clay Shirky makes them out to be in that article.
posted by Hildago at 7:21 PM on February 21, 2006

In the Ontology is Overrated article, Shirky makes a really basic mistake about indexing. His big example is this:

Let's say I need every Web page with the word "obstreperous" and "Minnesota" in it. You can't ask a cataloguer in advance to say "Well, that's going to be a useful category, we should encode that in advance." Instead, what the cataloguer is going to say is, "Obstreperous plus Minnesota! Forget it, we're not going to optimize for one-offs like that." Google, on the other hand, says, "Who cares? We're not going to tell the user what to do, because the link structure is more complex than we can read, except in response to a user query."

Seems to make sense, right? Wrong. Shirky confuses indexing words with indexing concepts. Roget's thesaurus lists seventeen synonyms for "obstreperous," which is about normal. Shirky's correct in that an index does a poor job of free-text searching. That's sort of the point. What he misses is that "Let's say I need every Web page with the word "obstreperous" and "Minnesota" in it." is very different from "Let's say I need every Web page about both obstreperousness and Minnesota or the ways in which Minnesota has demonstrated the quality of being obstreperous."

Shirky's point about biases in indexing is accurate, if basic. I mean, the point of classifiying things is to make judgements--judgements about what things are and what they're about and how they're related to each other. This is where a classification gets it's strength. If one removes the discriminating power of a classification, you get a random list of resources and have to go through them manually. The idea of a good classification is that it matches the perceptions of the user and delivers results that the user finds relevant. Folksonomy are, as Hildago says, great for a single user. Who, after all, knows how you think about things better than yourself? Of course, this means that the amateur user Shirky writes about is, in fact, the greatest subject specialist of all time. Why? Because each user has spent literally their entire lives performing an introspective user analysis.

Does that translate to other users? Sometimes. Some of my favorite systems combine free-text search, controlled vocabularies, and "tags". Multiple access points are great and the more we can offer the user (without presenting them with a screen full of meaningless noise) the better. It's that parenthetical that's the problem.

Or, more briefly, folksonomies are the ultimate manifestation of use warrant, but use warrant alone doesn't make a good index.
posted by stet at 9:42 AM on February 22, 2006