I was heavily involved in semantic web, ontologies, and natural language processing in my last company. I was supposed to work on that in my current company. In other words, I’ve been following this stuff for a while. The realization I came to is that, while it would be cool if everyone added all this metadata to their content on the Internet, and the cylons could take care of complex tasks for us automatically, like “Book me a room in a hotel near the Hynes Convention Center next Friday through Sunday with one queen size bed, bill it to my Visa card, and send the confirmation email to my Blackberry”, it’s not going to happen. The simple truth is it’s a lot more work than most content creators think is worthwhile, and adding metadata to existing content is nigh impossible. Just to save some businessman about an hour poking around the Internet for hotel recommendations and booking a room.I haven’t seen a lot of articles agreeing with me, but then I haven’t seen a lot of articles showing real-world examples of the semantic web working in some practical way (I’m sure you can guess why). However, an article was posted on LinkedIn‘s Semantic Web Group which showed more clearly exactly what I’m talking about.
The Gospel of Matthew says, “by their fruits ye shall know them.” Judging by their work, some of the biggest Semantic Web proponents are snake-oil salesmen. What else are we to conclude when academics and industry figures who fervently boost the Semantic Web can’t be bothered – or are unable – to publish their own Web materials with semantic mark-up? The day they get their own acts together, that’s the day the Semantic Web will emerge as more than just a questionable, always-just-over-the-horizon panacea for whatever ails Web users, as more than a justification for academic conference junketing, to solve real-world information findability problems.
When I moved from my own home-built-from-scratch PHP website to WordPress a few months ago, the main reason I did so is to more easily include metadata in and about my posts, automatic RSS feeds by tags and categories, and to allow others to do add their own metatadata to my posts by way of comments and ratings. That’s why so many of my posts are just a paragraph or two littered with hyperlinks to others’ content and heavily tagged. The lesson I learned is that it is possible to manage that kind of metadata efficiently, given the right tools and an early start.
The author of this article (Seth Grimes) makes another good point, though. Even if it were easy for companies to post RDF/OWL/Whatever information about their business, it might not be in their best interest to do so, because that information might also be accessible by their competitors. With so many commodity markets out there that compete mostly on price, one can only imagine the chaos that would ensue if companies set up autobots to make their prices 2% cheaper than the other guy. Exhibit A: The stock market.