Forgot your password?
Close
typodupeerror
Data Storage

Journal perfessor multigeek's Journal: Subdividing The WORLD 26

Journal by perfessor multigeek
Clearly, I'm not being clear.

*sigh*

Okay, instead of describing the category selections, hows about I show you?

So, all of my databases have a field called Category . This enables sorting records and searches so that, unlike googling, you can avoid false hits if you're looking for something. If you want to know where to find surplus computers for sale, you can search the records for ones with the category entry Shopping.Tech.Surplus. Look at the bottom of this entry to see what I mean.

Now such a system has several obvious shortcomings.
First of all, there are literally billions of possible selections. No matter how specific, somebody somewhere will eventually want to get more specific. Someday, somebody will insist that there should be a Shopping.Tech.CRT.Vector.Surplus.Military.Funky.
This is why I said in the previous JE that eventually the popup will have to have enough intelligence to provide nested options instead of all at once.

An even worse shortcoming is that most worthwhile sources (be they links, books, or whatever) fit into more then one category. You tell me, which is "the" "correct" category for Slashdot? For now I am using the extremely primitive solution of having multiple Category fields so a given source can be both Biz.Operations and Content.Commentary, as well as SciTech.OS.Linux. I made this choice so that first round distribution by tab-delimited file or DBF or SLYK would work better.
Eventually Category will have to be an array with sliders for each choice so a source can be listed under many categories and rated Very Strong to Very Weak on each. My suspicion is that a maximum of about twenty categories will be best to prevent reviewer laziness. That and a.) the forced selection of one to three "prime" categories and b.) a keyword field as backup should make the system pretty much bulletproof.
Keep in mind, by the way, that my system *does* include a text description field and is perfectly capable of being accessed with a search engine. I'm not religiously opposed to searching, I just don't consider it enough.

My general philosophy, decided in '97, has been that the SQL-derived engines are on the way to becoming universally applicable with many people able to create SQL work. But in the meantime a.) I don't have the time to sit down and learn the stuff, b.) the flavor wars are well under way (My- vs. MS- vs. Postgres- vs. Zope vs. ...), and c.) several basic kinds of functionality (e.g. transactions) are still being debated.
That all being the case, I decided to work as low-tech as I could (keep in mind that after years of high-end production for a living, I do most of my own site in Simpletext) and focus on the language and structure problems. What fields? What field types? What entry screens?, etc.
I concluded that I would blithely do things "wrong" as long as that "wrong choice" eased work now and could later be converted to the "right choice" with a script.

So, without further ado, I give you the list of my current Category choices.

Rustin

- - - - - - - -Start - - - - - -

Reed&Wright
Arts.Conceptual Arts.Fashion
Arts.Humor
Arts.Language
Arts.Language.Fiction
Arts.Language.Publishing
Arts.Movies
Arts.Music
Arts.Other
Arts.Visual
Biz.Auctions
Biz.Bookstore
Biz.Corporations
Biz.IntellectualProperty
Biz.Job Hunting
Biz.Marketing
Biz.Operations
Biz.Publishing&ContentDistribution
Biz.Research
Biz.Startup Stuff
Biz.Venture capital
Biz.Web
Content.Academic
Content.Comics&Graphic Novels
Content.Commentary
Content.Education
Content.Genre.Fantasy
Content.Genre.Mysteries
Content.Genre.Romance
Content.Genre.SpeculativeFiction
Content.Genre.Thriller
Content.Humor
Content.Images
Content.Images.Erotic
Content.Music
Content.News
Content.News.Local
Content.News.WebIndustry
Content.Paintings
Content.Publishers
Content.Pubs.Alternative Newspapers
Content.Pubs.Art&Design
Content.Pubs.General
Content.Pubs.Language
Content.Text.Fiction
Content.Text.Philosophy
Content.Text.Sacred Works
Content.Text.Poetry
Content.Pubs.Politics
Content.Pubs.Tech
Content.Zines/Personal
Design.Architecture
Design.Advertising
Design.Clothing
Design.Images
Design.Industrial
Design.Interior&Landscape
Design.Materials
Design.Microcars
Design.Publishing
Design.Type
Design.Web Site Creation
Design.Web.Fonts,Images&Sounds
Education.All
Education.Cognitive
Education.Tech
Entertainment.Games
Entertainment.Movies
Entertainment.Music
Entertainment.Radio
Entertainment.Televison
Entertainment.Theater
Entertainment.Web
Epicure.Consumer
Epicure.Food
Epicure.Household
Epicure.Novelty
History.All
History.Art&Design
History.Military
History.Politics
History.Tech
Homelife
Homelife.Finances
Homelife.Food
Humor.Dot Comedy
Humor.Offbeat
Humor.Personal Life
Other
Place.MetroArea.Amsterdam
Place.MetroArea.AustinTX
Place.MetroArea.MissoulaMT
Place.MetroArea.New York City
Place.Nation.Afganistan
Place.Nation.China
Place.Nation.Guyana
Place.Nation.Holland
Place.Nation.India
Place.Nation.Iraq
Place.Nation.Japan
Place.Nation.Mexico
Place.Nation.Phillipines
Place.Nation.Russia
Place.Nation.Samoa
Place.Nation.Tonga
Place.Nation.US
Place.Nation.Vietnam
Place.Region.Asia, Central
Place.Region.Asia, Southeast
Place.Region.Pacific Islands
Place.State.California
Place.State.Montana
Place.State.New York
Place.State.Pennsylvania
Place.State.Vermont
Reference.Geography
Research.Commerce
Research.General
Research.Technology
SciTech.All
SciTech.Aerospace
SciTech.Automotive
SciTech.Bio.Agriculture
SciTech.Bio.Cloning
SciTech.Bio.Reptiles
SciTech.Chemical
SciTech.Cooling&Refrigeration
SciTech.Construction
SciTech.Crypto&Security
SciTech.Desktop Hardware
SciTech.Desktop Software
SciTech.Digital Asset Management
SciTech.Electrical
SciTech.Elec.LEDs
SciTech.Electronics.Computer
SciTech.Fictional
SciTech.General
SciTech.Green.Compost
SciTech.Green.Greenmantle
SciTech.Green.Organic Farming
SciTech.Green.Other
SciTech.Green.Permaculture
SciTech.Green.Power Generation
SciTech.Green.Recycling
SciTech.Green.Water Processing
SciTech.Household
SciTech.Industrial/Manufacturing
SciTech.Infrastructure
SciTech.InterfaceDesign/Usability
SciTech.Lightweight Concrete
SciTech.Logistics
SciTech.Macintosh
SciTech.Maritime
SciTech.Material.Construction
SciTech.Materials
SciTech.Mathematics
SciTech.Mechanical
SciTech.Mech.Vehicles.Air
SciTech.Mech.Vehicles.Ground
SciTech.Mech.Vehicles.Water
SciTech.Medicine
SciTech.Med.Aging
SciTech.Med.Allergies
SciTech.Med.Cancer
SciTech.Med.ImmuneDisorders
SciTech.Med.Krone's
SciTech.Med.Nutrition
SciTech.Med.Parkinsons
SciTech.Med.Respiratory
SciTech.Micropayments
SciTech.Military
SciTech.Motor/Engine
SciTech.OEMs
SciTech.OS.Macintosh
SciTech.OS.Others
SciTech.OS.Palm
SciTech.OS.Unix and Linux
SciTech.OS.Windows
SciTech.Off the Grid
SciTech.Open Source
SciTech.Optics/Photography
SciTech.Other
SciTech.PDAs
SciTech.PDA.old
SciTech.Physics
SciTech.Publishing
SciTech.Recyling&Reuse
SciTech.Retail
SciTech.Robotics
SciTech.Shopping
SciTech.Simulation
SciTech.Software
SciTech.Software.Database
SciTech.Software.Database.SQL
SciTech.Toys
SciTech.Transportation
SciTech.Weapons
SciTech.Web
SciTech.Windoze/Linux hardware
SciTech.Workflow
SciTech.XML
Services.Hosting
Services.Printing.Poster
Shopping.All
Shopping.Art and Design
Shopping.Auctions
Shopping.Body/Mind
Shopping.Books
Shopping.Business
Shopping.Clothes
Shopping.Clothes.Radical
Shopping.Country/Camping
Shopping.Garden
Shopping.Green
Shopping.Health
Shopping.Housebuilding
Shopping.Housewares
Shopping.Jewelry&Other Body Ornament
Shopping.Maritime
Shopping.Materials
Shopping.Materials.Fabric
Shopping.Office Stuff
Shopping.OfftheGrid
Shopping.Surplus
Shopping.Tech
Shopping.Tech.PDA
Shopping.Tech.Surplus
Shopping.Toys
Soc.Advertising
Soc.Arts.SpeculativeFiction
Soc.Beliefs&Ideologies
Soc.Biography
Soc.Business.VentureCapital
Soc.Commerce
Soc.Crime
Soc.CrisisResponse
Soc.Death
Soc.EconomicsTheory
Soc.Elections
Soc.Ethnicity
Soc.GenderIssues
Soc.HistoricRecreationists
Soc.Identity
Soc.Identity.Christian
Soc.Identity.Geek
Soc.Identity.Goth
Soc.Identity.Hippie/Raver
Soc.Innovation
Soc.Language
Soc.Law
Soc.Law.IntellectualProperty
Soc.LawEnforcement
Soc.Military
Soc.Military.Protocols
Soc.Org
Soc.Org.Aerospace
Soc.Org.Corporations
Soc.Org.Government
Soc.Org.Intentional Communities
Soc.Org.Interest groups
Soc.Org.Legislatures
Soc.Org.Military
Soc.Org.Nation
Soc.Org.Nation.NativeAmerican
Soc.Org.Researchers
Soc.Org.Schools
Soc.Organizations
Soc.Organizations.Business
Soc.Politics.All
Soc.Politics.Bush Jr. Administration
Soc.Politics.Censorship
Soc.Politics.Clinton Administration
Soc.Politics.Content&Free Speech
Soc.Politics.Drugs.General
Soc.Politics.Drugs.Marijuana
Soc.Politics.Green Issues
Soc.Politics.Law Enforcement
Soc.Politics.Open Source
Soc.Politics.Prisons
Soc.Politics.Reagan Administration
Soc.Politics.Warfare
Soc.Politics.9/11 Response
Soc.PopularCulture
Soc.Poverty
Soc.PovertyReduction
Soc.Psychology
Soc.Punishment
Soc.Scitech.cloning
Soc.Scitech.General
Soc.Sex
Soc.Sports
Soc.Violence
Soc.Warfare
Soc.Web
Tools.Filmmaking

- - - - - - -End - - - - - - -
This discussion has been archived. No new comments can be posted.

Subdividing The WORLD More

Subdividing The WORLD

Comments Filter:
  • Ohh, it's Usenet with caps and no spam. ;-)
    • Ohh, it's Usenet with caps and no spam. ;-)
      That is not a bad comparison. I have long suspected that anything that merits a group on Usenet should qualify as a category. After all, the acid tests should be: Is this
      a.) something that interests people?
      b.) a degree of subdivision/specificity useful to those people
      c.) matching the cognitive mapping that users already have for this topic.?


      The existence of a Usenet group confirms the validity of all three.

      Of course, as the expression goes, to categorize
      • It all could get seriously complicated

        Hooboy, it sure can. I've designed and built a few semi-large intranets (one of which has the dubious honour of belonging to a company later bought and killed by Bill'n'Ballmer's Runaway Circusss) and categorizing content in meaningful ways was the most difficult task of them all. Most of the time, I cheat a little and make stuff available through at least two totally different viewpoints; like listed by which process the data belongs to and which type of document

  • Why are the hippies and ravers grouped together?
  • I encourage you to go forward with these categories. I don't have much more to say that than that. If I get a chance, then I look forward to browsing around.
  • i like what you are doing but have a suggestion for how the categories work.

    i think by imposing the tree structure, rather than letting the tree grow, you are going to have situations where information's definition within the limited confines of the existing categories is difficult/impossible.

    a simple example: content.images and design.images are both sites for images. many sites could easily exist in either and the choice becomes arbitrary. i suggest droping the imposed heirachy -- instead define d
    • Again I was vague. I was afraid that I needed to clarify that one.
      All that I intended to say was that each category should have an associated slider with which one could determine degree of applicability. Certainly all Categories overall should not need to "add up to 100 percent" though a 1 to 100 range makes sense for each cat. in isolation.
      That was why I used a Strong/Weak continuum rather than a numerical value. While the stored rating would be numerical, decoupling from a cumulative or numerical way of
    • i've messed around with content management myself but never had enough content to make it fun. maybe i should just ask for your data and start hacking....

      And now you are doomed indeed!
      No sooner suggested than done. I'm mailing you DBFs of the biographical and chronological DBs today, just for starters.

      Doomed, I say, Doomed!
      Mu-hahaha!

      Rustin
    • Oh gawd, I just had an evil thought.

      How would one work with the concept of category proximity vs. category similarity?

      IOW, it would make a lot of sense, once there were enough entries to make it interesting and the cat system is richer, to index a given cat relative to other cats. Which ones does it tend to share entries with? Do sources with Epicure.Food.Vietnamese tend to also have SciTech.OS.Linux.Debian?

      This should lead to the ability to derive meta-categories, in this case having sources including
      • I'm starting to visualize this as a honkin' big sphere full of floating dots, each dot being a quanta of data. The categories are coloured volumes in this vast space, enclosing dots and sometimes (often) overlapping. Only problem is that 3D isn't really enough to connect the dots, as it were. This universe really needs wormholes...

        Well, as long as it doesn't give "A suffusion of yellow" [thateden.co.uk] as the answer to too many queries, it should be fine. ;-)

        What would be needed, hardware- and software-wise to get this

        • What would be needed, hardware- and software-wise to get this thing online?
          Well, for the moment, since it's so tiny and dinky (fewer then two thousand entries in the largest DB, all flat file, only two large fields with the largest field size under two hundred characters) I could host it on a Mac 8100 and be fine. Fer cryin' loud, at this point it could probably be hosted on a Newton.
          The question, however, is what would it take to bring the thing on line in a usable form with some relations in place.
          Wel
          • Here some random for ya. Use bellus quies' "Specific.Question.Why?".

            Have that as a category and when selected, kicks out a random record from somewhere in the DB. Kinda answers itself in that case. Be interesting to track what came out.
          • SQL is a bear for indexed searches.

            I'm no db expert, but it seems to me that indexed search performance is not a SQL problem per se, but an implementation problem. SQL is a bit too standardized to skip... AFAIK, Slashdot's search problems aren't related to indexing but free-text and in GM, free-text searches should become secondary (and we could always googlify that specific functionality, right?) to the cat searches which should be a lot cheaper, resource-wise.

            I was thinking that instead of just talk

            • I'm no db expert, but it seems to me that indexed search performance is not a SQL problem per se, but an implementation problem. SQL is a bit too standardized to skip... AFAIK, Slashdot's search problems aren't related to indexing but free-text and in GM, free-text searches should become secondary (and we could always googlify that specific functionality, right?) to the cat searches which should be a lot cheaper, resource-wise.
              Cool. That all makes sense to me. I'm glad to hear that I was wrong. As I have
              • I'm glad to hear that I was wrong.

                You didn't read this claimer: "I'm no db expert". That said (again), I still think indexed search performance is a non-problem since that's a main reason why we have indicies in the first place - to speed up searches and selections on them.

                an Arthur C. Clarke book with a solar-system-wide DB

                Is that the short story where they are finally down to quarks holding lots of info and still the entire solar system is taken up by the index and cross-references to the library

  • The name for this system comes, regularly as clockwork on all-bran, from the Daily Dilbert (I never read my horoscope, Scott Adams is my fortune-teller):

    Geneous Mistro [dilbert.com].

    • That's it. Geneous Mistro it is. Now we jst need to get Scott Adam's permission.
      Rustin
      • I'll bet that you thought that I was kidding.

        I just sent the following email to Scott Adams:

        - - - - - - - letter starts - - - - - - -

        Mr. Adams, I have been working on an information management system for six years and have just started the process of open-sourcing it this week. As a former QPS admin and content managment guy for folks like Time-Warner, McGraw-Hill, and J.Crew, I take this project very seriously, but never solemnly.
        The following thread on Slashdot should give you a sense of what we'r
        • Wouldja believe that I just got the following email?

          - - - - - - letter starts - - - - -

          Hi Rustin,

          You may use the Geneous Mistro name.
          Unless someone else already has it.

          Thanks for asking.

          Scott

          - - - - - - letter ends - - - - -

          Well then. Geneous Mistro it is.

          Can I just say that I love living in a time when I mail an unsolicited question to a national figure on a Saturday afternoon and an hour and a half later he's responded and said yes?

          I LOVE LIVING IN THE FUTURE!

          Now we just need a t-s
          • So, tell me what you were thinking. I also need to go over the LSC logo, for a new SLED (tm) project, SuperPunk (tm).

            Flash is the coolest.

            • Howsabout "Geneous Mistro" in big bold sans serif letters with an opaque stroke but a fill of Courier URLs, magazine and book titles, names of political groups, etc.?
              Of course, I'm too lazy to sit down in Illustrator and actually do the work (most of which would, I suspect, be determining optimum stroke weight and making the type itty-bitty but pretty).
              You are certainly more then welcome to wade in. After all, you did hundreds of entries into the beast. You may, in fact, have spent more hours doing entries

Slashdot Top Deals

The decision doesn't have to be logical; it was unanimous.

Close