The importance of precision in markup

Here is hoping you had a safe and happy holiday season. During the holidays, I had an experience in buying a present that made me think about the importance of markup, and more importantly the necessity to be precise when dealing with markup.

My brother-in-law, like me, is a big Boston Red Sox fan, so I wanted to give him some kind of item with the team logo on it, but something a bit different than a T-shirt. A few searches and I finally found the perfect item, a set of Red Sox logo slippers on Amazon. I made the purchase, but then was surprised to see the following (image below) when I checked out. You may have to expand the image to see it, but there it is, plain as day. People who bought this item (Boston Red Sox slippers) also bought the following item (New York Yankees slippers).

Excuse me?

For those of you who are not baseball fans, these two teams -- and more importantly, their respective fans -- hate each other. The very idea of a Red Sox fan also buying anything embossed with the Yankees' logo, or vice versa, is ridiculous. It would be like a member of the Hatfields ordering a McCoy's T-shirt, or someone from North Carolina ordering a pair of Duke slippers. Just ain't gonna happen.


So why was this an option on the Amazon site? Clearly the products are similar on one level (ie, they are both slippers), and perhaps someone buying every slipper for every Major League Baseball team would want to know about the other teams available, but is this useful information for a Red Sox fan? I think not.

I should note here that I am aware of *how* this information got there. The logic on the Amazon site is pretty machine-driven, and clearly someone purchased both of these slippers, so the result is technically valid. But there's a world of difference between markup that is technically correct and markup that is useful (it's roughly the same as the distance between Fenway Park and Yankee Stadium).

That made me think about markup, and the *purpose* of markup. Serving up different "planes" of knowledge can be accomplished by machine pretty easily. Take a hammer. If the system shows a hammer, that's a tool, so it shows all other tools. That can be pretty useful to someone performing a search on different types of tools. But one of the main benefits of semantic markup is to improve the meaning, and precision, of searching and finding. "If you searched for *this*, you might also be interested in *that.* A noble goal, but it is not as simple as it may seem.

Enter intelligent, precise, semantic markup.

Surely there is no harm in the Amazon example I've cited -- I can simply chose to not order an item with the Yankees' logo. But when doing a search or browsing for information of a more academic nature -- for example, professional medical content -- the serving up of related items that are related but not useful, can, at best, mislead and, at worst, mis-inform.

I am reminded of the quote attributed to physicist Richard Feynman, who said: "Philosophy of science is about as useful to scientists as ornithology is to birds." Items served up by semantics should not just be related on any plane, but the connections they enable should be useful (certainly not in all cases, but the connections should be discernible by a reasonably intelligent person, and most certainly not cause confusion). This is a tall order, but can be accomplished by the intelligent insertion of semantic data by subject-matter experts (read: humans) who can ask and answer the important questions about the content, and how it will be consumed.

--Jabin White