The Semantic Web and the I Love You problem

Since I started out with a nod to Ben Hammersley, I’ll risk biting the hand that inspires me by taking issue with something he just posted to his blog.

It’s a long-lost (and nicely produced) video of an interesting and accessible talk he gave in summer 2003 entitled A Sporting Gentleman’s Guide to the Semantic Web. It seems like a useful intro for someone like me who knows very little about the subject.

However, despite a promise not to oversell the Semantic Web, Hammersley may be doing just that. He claims that it solves what he calls the “I Love You” problem, that the same words can mean different things when said by different people in different contexts. The Semantic Web aims to address this in two ways: (1) by using unique identifiers, or URIs, for each of the terms in a statement; and (2) by defining its verbs via an explicit reference to a standard.

So far so good, but the example he uses to illustrate his point is one that sets off alarms for me, the Dublin Core element dc:creator. It would be hard to name a thornier category of metadata. Any cataloger will tell you that there are a thousand kinds of authorship. The Dublin Core definition he refers to in his talk does not make the concept unambiguous, but rather explicitly ambiguous:

Creator: An entity primarily responsible for making the content of the resource. Examples of a Creator include a person, an organisation,or a service. Typically, the name of a Creator should be used to indicate the entity.

So neither the Semantic Web nor the Dublin Core will tell you whether the dc:creator of the object conceived it, wrote it, directed it, produced it, acted the lead, played trombone in the pit, or did the catering. They say even less about what rights or responsibilities accrue to said creator. Those questions remain as messy with the Semantic Web (at least the Dublin Core-based version) as without it.

Probably I’m overreacting and the problem isn’t with the Semantic Web but with his example. More widespread machine-parseable use of a simple metadata standard like Dublin Core could come in handy despite its limitations, and if I understood it correctly the Semantic Web is extensible to any metadata standard, including much more precise ones. But a claim that the Semantic Web eliminates issues of ambiguity is an overstatement as far-fetched as the claims of magic and mind reading his talk is intended to debunk.

3 thoughts on “The Semantic Web and the I Love You problem

  1. Very good point. I think it’s a bad example, rather than a just bad logic. I chose dc:creator because it’s a simple example that everyone in that audience could relate to (they were all either coders or artists, so the concept of a single creator of a document was natural to them). But, as you say, dc:creator is a tad too ambiguous in real life. I should have used a date value or continued with the Pico example.

    Still, 18 months on, it all vaguely holds up, which is rather surprising.

  2. Fair enough.

    I’m pleasantly surprised to see your comment here — I thought I was operating this blog in semi-stealth mode while I tinker with it and get some content together. You’re right behind the spambots in finding me out. Perhaps WordPress has trackback/pingback features enabled by default that escaped my notice. Or perhaps you just read your referrer logs.

    Of course if I’d been serious about flying under the radar I’d have bothered to put in a .htaccess file.

Leave a Reply