Melody Kramer

How Many Ways Could We Categorize This Piece

09 Mar 2015

Today’s newsletter! (Want to subscribe? Click here.)

I picked a random piece from the NPR archives. Let’s think of all of the different ways we could categorize this piece.

  1. About: Fiesta
  2. About: pottery
  3. Time: under 8:00 minutes
  4. Time: over 7:00 minutes
  5. Author: by Linda Wertheimer
  6. Author: female
  7. Subject: brothers
  8. Subject: history
  9. Place: West Virginia
  10. Subject: labor
  11. Subject: manufacturing
  12. Date: October 2014
  13. Sentiment: thoughtful
  14. Categorization: new facts about something you may have heard about
  15. Categorization: a human interest story about an American product
  16. Categorization: a human interest story about West Virginia
  17. Categorization: pieces that contain a secret
  18. Categorization: pieces for people who may want to learn more about dinnerware
  19. ” “ … about West Virginia
  20. ” “ … about an American product
  21. ” “ … about Homer Laughlin
  22. ” “ … people who collect Fiesta
  23. ” “ ….collectors
  24. Feeling: hopeful
  25. Feeling: optimistic
  26. Feeling: cheerful
  27. Contains: person who works on manufacturing line
  28. Contains: person who works in a museum
  29. Contains: person who works in a family-owned business
  30. Illustration: animated gif
  31. When to listen: in the car with your kids
  32. When to listen: after a newsier piece
  33. Categorization: companies with more than 1,000 employees
  34. Categorization: family-owned businesses
  35. Place: Newell, West Virginia
  36. Show: Morning Edition
  37. Evergreen: yes

That’s all I got. What am I missing?



From Bryant Welch:

Host: Rachel Martin Series: American Made: The New Manufacturing Landscape ( Photographer: Ross Mantle Season: Fall 2014 Geographic reference: Ohio River Geographic reference: Japan Geographic reference: Mexico Geographic reference: Boston, MA

From Ted Coltman

Very interesting exercise. This may be too much detail, but two additional aspects we could categorize are the business economics embedded in this piece and more info about the illustrations accompanying the piece on its archive webpage:

Subject: ‘retro’ revival of consumer demand for formerly obsolete products Subject: manufacturing products for dual retail and institutional marketing channels Subject: manufacturing products for multiple countercyclical marketing channels Illustration: 14 JPEG still photos and 2 animated GIF photos Illustration: 2 animated GIFs of production machinery, with products moving along production line Illustration: still photos of 2 male and 2 female employees, other members of whose families have worked for same company etc.

From Sonya Mann

Companies 100+ years old. Americana. Crafts. Generational work.

From Gabe Isman

When I see this list the part of my brain that has done data modeling starts complaining loudly. “Melody, you shouldn’t be trying to make the biggest list, you should make the smallest list that doesn’t lose anything!”

So to my mind, 4 and 5 are totally redundant, that the piece is 7:22 should be enough. Likewise 5 and 6, 9 and 35, etc.

I know these are really something like ‘lists this piece could be long to’ and not ‘data about this piece’, so my data modeling brain should just shut up. But I think something interesting is highlighted by picking out which of these categories is somehow essential, i.e. couldn’t be deleted without loss of information. Some surprisingly obscure bits become essential, for example:

  1. Categorization: new facts about something you may have heard about
  2. Categorization: pieces that contain a secret
  3. Categorization: pieces for people who may want to learn more about dinnerware
  4. Feeling: cheerful

Not surprisingly, I suppose, these are also things that are hard for a computer to figure out on its own. These bits of data are somehow more human than some of the others.

From Kate Myers

First – fantastic piece! You know this, but your newsletter is a really nice bite of insight. I admire you for doing it.

My 2 cents to add - Cult icon, something that people collect.

Then you can take a step to communities that would engage with this – those collectors, people who are interested in the backstory of classic americana, etc.

If we were going to recast all these data points from the perspective of the audience, what would they be? You’re on to something with the “when to listen”, but I wonder if there are more.

From Andrew Losowsky

Decibel rating Waveform Show producer Times broadcast Stations on which it was broadcast

And your own listener data: how you found it, when you found it, had you listened to it before, did you listen to it all, what rating do you give it, who if anyone would you tag it with to share.

From Daniel Newman

The ‘Categorization’, ‘Contains’, & ‘When to Listen’ pieces seem most useful, but also the hardest to systematize. Figuring out how to represent the ‘shades of evergreen’ would be one of the more useful things, if that could be done.

From Sue Diaz

Is ‘subject’ a fixed taxonomy? I would add the woman-led biz aspect, maybe in ‘contains.’

From Josh Tong

Great list! Have you seen the Dublin Core metadata terms? Could include format (audio, text) and rights.

Leave a comment
Fix my typos