If the Semantic Web depends on the availability of massive amounts of high-quality metadata
Then it won’t and can’t possibly work.
You know why?
BECAUSE PEOPLE ARE MOTHERFUCKING RETARDED MORONS AND COULDN’T PRODUCE QUALITY DATA IF THEIR LIVES FUCKING DEPENDED ON IT.
And all the propellerhead articles and theories and software and other bullshit in the world won’t change that fact.
Have a look at any publicly available set of data and try getting quality without paying through the fucking nose. Good luck, dumbass.
No, I’m not frustrated by my job at all, why do you ask?
The nice thing about the web is that there is just so much information that the bad data averages out over a long period of time. Think of wikipedia, all the info in there is data. If we were to somehow parse the unstructured text and derive semantic relationships from it, we will have ourselves a powerful semantic web. Ignore this parsing for now, consider making a participatory semantic web along the lines of wikipedia except that in this semanticPedia entries must have some meta-data associated with them and their links between them are well-defined. Of course, there will be many among us who will get it wrong, of course there will be attempts to sabotage the effort by putting wrong meta-data but so what. These were the same arguments used against wikipedia, remember. And now wikipedia boasts the largest collection of encyclopedic information on the web.
I do agree, however, that if semantic web was to be used for commerce, the accuracy requirements would be much more stringent than wikipedia. It’ll still be awesome for significantly our searching algorithms and the like as we can tolerate a few bogus search results.
Comment by khurramm — May 15, 2007 @ 7:06 pm
People aren’t bad at recording knowledge. The problem is that most metadata (really just structured data, in most cases) is too abstract for most people, and frequently the structures that have been devised are too limiting. On top of that, most structured data is usually quite subjective — well, the useful stuff is, anyway.
The semantic web will certainly never work until computers can understand human languages. Or the data is generated as a side effect of doing something good — datestamps are a good example. But that sort of data is boring.
I’m not sure what your job is, but if it’s e-commerce related, then that’s even more difficult. There really needs to be several layers of actual metadata to go with the data; systems need to actually _understand_ stuff, rather than just recording it and regurgitating it at the right time.
Not that this comment helps your mood, I’m sure…
Comment by simon — May 16, 2007 @ 12:57 am
[...] Cory Doctorow outlines some techniques for dealing with bad characters in your online community. The semantic web is doomed I have similar frustrations. I don’t state them quite this colorfully, [...]
Pingback by » Linky Goodness - 5/20/2007 — May 20, 2007 @ 8:50 am