I came across some interesting notes while going through old emails the other day. A message from NISO, the National Information Standards Organization, reported that the semantic web is dead, citing a post on semantico. The semantic web is a concept of presenting data in a structured format, usually as ‘triples’ (I am, absolutely, not an expert – or even that knowledgeable – on this stuff, so don’t quote me too far), so a computer can better understand what each term means.

For example, when a computer sees the word “Magellan”, it just sees a word. It doesn’t know if the word refers to an explorer, to a spacecraft, to a mutual fund, a “progressive metal/rock” band, or something else. By defining, through triples, what one means, the computer can realize that one page is talking about the explorer while another is talking about a mutual fund company.

Such semantic definitions have been used extensively in some subject areas, but not at all in most. And one of the great challenges with it is/was solving problems among the “upper ontology” – that is, the layer that connects concepts in zoology with concepts in art history with concepts in electrical engineering with concepts in maritime history, etc. One field may work hard to define its ontology, but if that schema doesn’t mesh with other ontologies, then the systems aren’t really connected.

So I was interested to read of the effective death of the semantic web, and its replacement by schema.org. Schema.org is a nascent project being put together by representatives from the search teams at Google, Yahoo, and Microsoft’s Bing. It uses microformat HTML tags, added to a page’s markup text, to define what something is. This is done for the benefit of search engines – so a “Magellan” that is marked with the tags

  <div itemscope itemtype="http://schema.org/Person">
    <span itemprop="name">Ferdinand Magellan</span>

is clearly a person, while the Magellan that’s tagged

  <div itemscope itemtype="http://schema.org/Product">
    <span itemprop="name">Fidelity Magellan Fund>/span>

is something you can buy. (Note the differences in the end of each first line; the first is “/Person”, and the second is “/Product”.)

(Also: I defined the Magellan Fund as a ‘product’, because one can buy a share of it, but it might more appropriately be an ‘organization’, since there is a ticker symbol associated with it, and schema.org currently has a “tickerSymbol” attribute for Organizations.)

The current schema.org structure is quite limited, and focuses primarily on people, organizations (especially local businesses), creative works, events, and locations. But it’s certainly extensible, and – if it’s generally adopted, as triples were not – it will clearly expand to other fields.

I’d love to take on extending it to vessels. It’d be pretty easy for us to modify our HTML to include these microtags, and if that helps people find the information they’re seeking, then all the better for all involved. But I’m not sure what the proper levels should be. One doesn’t want to have too many levels in a structure like this, but I think that going straight from “Thing” to “Vessel” might be a bit of a jump. I imagine an intermediate step of, perhaps, “Vehicle”, would be appropriate. Then those with interest in cars, trains, airplanes, bicycles, scooters and lots more, would build out their schemas, while we could start a layout of sailing vessels.

It seems simple, but immediately becomes fairly complex. You could, for instance, split up “Vessel” entries to “HumanPowered”, “WindPowered”, and “MechanicallyPowered”, perhaps, then divide by vessel type – canoe, kayak, paddleboat; sloop, ketch, yawl, schooner, brig, brigantine, barkentine, ship, bark, hermaphrodite brig; paddlewheel steamer, ferryboat, fishing boat, battleship, oceanliner; etc., etc. Is that too much differentiation? How do you define a vessel that’s been re-rigged, from a ship to a bark, for example? How, even, do you make it clear that when you’re talking about a ‘ship,’ you’re talking about a three-masted vessel with square sails on the furthest-aft mast, rather than something that floats and is bigger than a boat?

Lots of other terms could be added or defined over time. When the computer can understand what the term means, rather than just presenting the term to the world, it will make it much easier for individuals to draw understanding and make connections from within large bodies of marked-up data.

It would appear that this system, because it’s fairly easily applied, has a much better chance of success than did the original ‘triples’ approach. I look forward to watching it with interest.