Comments

4 responses to “Muffett’s rant on why The Semantic Web will never happen…”

1. 2007/05/04

   Geoff Arnold

   re: Muffett’s rant on why The Semantic Web will never happen…

   I see no conflict whatsoever – it’s a false dichotomy.

   The way “that which does the understanding” (presumably software) will “become more clever” is through interactions with other “understanding” systems, including humans.

   We already “describve” documents, every time someone assigns a Dewey Decimal or Library of Congress category to a book. My brother (who works in a library) would be delighted if a chunk o’ software would process each book and suggest a category; initially he would want to review and correct it as necessary, but as it became more reliable he would trust it more.

   Exactly how, in principle, is this different from tagging?

   Furthermore it would be extraordinarily inefficient if every piece of software that processed a document were to do so ab initio, starting with the quill-scratchings of the humans. Over time, “documents” will (should!) become composite entities with human-readab;le and machine-readable bits.

   Much of the work that I do at Amazon involves designing systems in which different types of information, with different rates of change, used for different purposes, should be stored, cached, joined, preprocessed for efficient use later on, and so forth. Very few of us are raggedy-bearded, although we are all mathematically-inclined…..

   Reply
2. 2007/05/05

   Stephen Usher

   re: Muffett’s rant on why The Semantic Web will never happen…

   Hmmm… I see a workload in classifying things becoming a gigantuan task, possibly even more so than writing the original matterial.

   As for getting “more clever” software to do the job for us, I don’t see that ever happening until we have fully sentient slave A.I.s which can parse all languages and read them and classify them as a native of the country the content came from would. i.e. the A.I. would need to be able to know context and cultural reference to understand the subtle inferences in the text. Oh, and then there’s the problem of tagging images… Now we get into robotic vision systems and spacial awareness. Nasty!

   All the “clever” software I’ve ever seen so far (except chess playing programs) seems only to make matters worse. If there are two choices invariably the wrong one is taken in my experience. I can’t see this field being any different.

   Sorry, I think the symantic web is just a pipedream and one of those “wouldn’t it be cool if…” thoughts which has been taken to the extreme and is now too high profile to quietly forget.

   Reply
3. 2007/05/05

   alecm

   re: Muffett’s rant on why The Semantic Web will never happen…

   >Exactly how, in principle, is this different from tagging?

   Intent. Librarians catalogue books within a fixed context for retreival, withh indices in a single language, and with a notion of how to approach lookup.

   The web is a librarian’s wild west. Few rules, and a lot of innovation amongst a lot more chaos.

   Reply
4. 2007/06/01

   Danny

   re: Muffett’s rant on why The Semantic Web will never happen…

   Note that the Semantic Web is about more than just describing documents – it’s about describing things, people, concepts, activities: anything which can be identified (with a URI).

   Ok, there are (human-readable) documents, but this is only a subset of the information currently expressed digitally. For example, there is already a huge amount of machine-readable data about the world locked up in databases, much of this can be usefully exposed on the web. Semantic Web technologies offer a common data model which works on the web, and there are a good number of tools which make this kind of thing possible and useful.

   I guess there are a few raggedy-bearded types around…but I assure you we know how to have fun 😉

   Reply