Join 3,417 readers in helping fund MetaFilter (Hide)


All your semantic web are belong to google
November 20, 2005 6:20 AM   Subscribe

All your semantic web are belong to google? The implications of Google Base on the fabled semantic web.
posted by MetaMonkey (25 comments total)

 
This is my first FPP, so please don’t be too upset by the numerous unspoken rules I’ve violated. I did try. Alternatively do be upset but be distracted by some crazy martial arts videos.

Let the flaying begin.
posted by MetaMonkey at 6:21 AM on November 20, 2005


You know, I've given this semantic web business a lot of thought and my main conclusion is that, man, Wing Chun Kung Fu is cool.
posted by nthdegx at 6:25 AM on November 20, 2005


Just what I was thinking
posted by MetaMonkey at 6:46 AM on November 20, 2005


This is exciting.
I'm waiting - patiently - for the day when my wearable interfaces directly with the Googlebot - and I never have to learn anything ever again.
All the information is just...there. I can shop and not worry about getting ripped off. I can quote media prices to my boss during the conference call.
The semantic web seems to be about information verification, fwiw. Validating snippets of info - or retrieving them through a sort of info-intuitive process. I'm all for it.
Hail Google - High-Lord of the Intertrons!
posted by Baby_Balrog at 6:47 AM on November 20, 2005


You have nothing to worry about. It's a fine post, especially since it isn't purely lip service to google.

Now, as regards the semantic web: I've always wanted the internet to function in this manner, because it'll be more like my favorite science fiction stories. 2 years ago, I would have been thrilled for google to lead the way. Now, I'm regarding google like that kid in class who aces all the tests, is football qb, heads 3 extra-curriculars, and is building a robot in his basement in his spare time (when he's not volunteering for the red cross). Something with the kid's just not normal!
posted by shmegegge at 6:49 AM on November 20, 2005


mostly because, in my favorite sci-fi stories, you'd then find out the KID'S the robot, and he's been building a real, flawed boy in his basement in his spare time.
posted by shmegegge at 6:51 AM on November 20, 2005


Excuse me, but "semantic web" and "Web 2.0"? Aren't these just buzzwords that dumb people use to sound important?</simpsons>
posted by eustacescrubb at 7:00 AM on November 20, 2005


"Aren't these just buzzwords that dumb people use to sound important?"

Not unless you count Tim Berners-Lee, inventor of this here web thingumy, and the semantic web, as dumb.

Pretty much spot on with the web 2.0 thing though, although from its nonsensical beginnings, it has somehow stumbled upon some kind of meaning.
posted by MetaMonkey at 7:30 AM on November 20, 2005


This is my first FPP, so please don’t be too upset by the numerous unspoken rules I’ve violated.

And then...

Alternatively do be upset but be distracted by some crazy martial arts videos.

Having just come from the legendary mushroom thread, may I propose that if only that FPP had had a link to crazy martial arts videos a lot of unpleasantness could have been avoided.
posted by Opposite George at 7:38 AM on November 20, 2005


"Aren't these just buzzwords that dumb people use to sound important?"
Not unless you count Tim Berners-Lee, inventor of this here web thingumy, and the semantic web, as dumb.


While Berners-Lee is no doubt a smart guy, that doesn't necessarily mean everyone who uses the term is smart as well. I personally have no idea what the 'semantic web' is supposed to be, so if I used the term it would just be a buzzword.

Now "web 2.0" that's definitely just something dumb people say to sound smart. Or at least hackneyed enough to get some VC funding.
posted by Paris Hilton at 7:59 AM on November 20, 2005


I see London, I see France...
posted by 517 at 8:14 AM on November 20, 2005


I used to understand the term semantic web as predicate logic in xml. Or, alternatively, as the second coming of AI.
The difficulty with conventional predicate logic is to come up with a single standard ontology that is really universal. This has been summed up in a very readable form by Clay Shirky
If the semantic web is really that powerful there should already be powerful examples of its use. Can anybody present some of these?
posted by jouke at 8:18 AM on November 20, 2005


To me, the semantic web is fundamentally about self describing data that is both machine readable and human readable. One of the nifty things about xml is that it not only presents data, it also presents meta-data, that is data about the data.

So if I have a recipe and I compose it in xml then I can have each ingredient in an ingredient tag. When a machine comes across it (or a human for that matter) and it sees all these entries each labeled "ingredient" then it knows they are ingredients. They might not really understand what an ingredient is, but at least they know that they are ingredients. If you just have an unordered list of things like "2 cups flour" "1 Tbs Applesauce" then the only way to know that those things are ingredients is context. I have to already know what a recipe is and what ingredients are and that these are examples of ingredients. And context is something that machines are not historically very good at understanding. Some would argue that people aren't necessarily very good at that either.

I would say that Metafilter's tagging system is a pretty decent example of the semantic web. The tags on this particular post are fairly straight forward and are all words that are contained in the post, so a decent search engine would pick up this fpp if I searched for "google". But some tags in the system are related to notions/feeling/context that are not explicitly found in the words on the page. Try looking at the "batshitinsane" tag. It gives you a thread of stories that fit a psychological profile. Few (if any) of those posts actually contain that word though. So it serves the function of letting me find things that are similar in a way the current search engines can't, because "batshitinsane" is metadata, not data :)
posted by afflatus at 8:42 AM on November 20, 2005


Google makes a habit of launching services amid fanfare and then letting them stagnate. Their primary search engine is the only thing being tended to properly, and even that has become plagued with issues (ranking sales sites above informational ones, etc.).
posted by fleener at 9:35 AM on November 20, 2005


the web was totally cool when it was one big graffiti wall.
posted by quonsar at 9:43 AM on November 20, 2005


I think the thing that annoys me the most about "folksonomies" is the confusion of lay people (and professionals!) around what users are actually doing when they "tag" images at flickr. They are not, as the Cnet article suggests, "classifying". The whole notion of the community of flickr users regulating a common semantic vocabulary ala Wikipedia is absurd. There is no more wisdom in the flickr crowd than in us collective google users and pagerank spammers, only a deep abiding interest in porn and other lowest common denominators. Even flickr's peer pressure technology threatens to get as ugly as that high school popularity contest you've been repressing all these years.
From Metametametadata. This blogger, I feel, expresses my feelings on folksonomies far better and more succinctly than I could do myself. The author's comments are every bit as relevant to Google Base as to Flickr.
posted by stet at 10:07 AM on November 20, 2005


I would say that Metafilter's tagging system is a pretty decent example of the semantic web.
Hm, a quick inspection learns that metafilter tags are simple keywords with duplicates (f.i. '80s', '80s', '80's') and no relations. That's okay, but it does not qualify 'semantic'.
If I'd equate 'semantic' with 'conveying meaning in a formal manner' metafilter tags are not 'semantic'.

I looked at Piggybank Firefox-plugin. What Mosaic did for the web, we think Piggy-Bank will do for the semantic web!. It's a firefox plugin that is RDF savy.
Has anybody experiences with that?
posted by jouke at 10:47 AM on November 20, 2005


By the way: Dr Youngs chi power is hilarious.
posted by jouke at 10:54 AM on November 20, 2005


See, I would have tagged this with 'ontology' ;)
posted by carter at 10:57 AM on November 20, 2005


jouke, thank you for the link to the Clay Shirky piece. I really enjoyed it and it clarified some of the things I had been thinking about with regard to the limitations of categories.
posted by beth at 11:12 AM on November 20, 2005


Object Oriented web-programming with dynamic object properties that cross-correlate?
posted by IronLizard at 11:47 AM on November 20, 2005


yeah, the semantic web for me means machines grokking stuff for you. Machines should be able to answer the question "What was the population of California in 1992?" and trillions of others.

If one can answer arbitrary questions correctly, one is at least useful, if not intelligent.
posted by Heywood Mogroot at 11:59 AM on November 20, 2005


Um. A buzzword can also have (had) an original, buzz-free origin. afflatus came closest to describing what the original meaning of "semantic web" was -- essentially data that could describe itself, that wasn't just a blob of data moved hither and yon.

Let me digress a moment and address stet's point about folksonomies. What I hope stet understands, and what I'm sure "Metametadata" understands, is that there are two competing philosophies at work in tagging, which are roughly classifying vs. labeling. Some people think they're doing one, others think they're doing the other. Which wins isn't the point, really -- it depends on what tools are available to leverage the tags in place. Here at MeFi (and a lot of other places) they're just disconnected labels. There isn't a Bush or Web "room" at MetaFilter, nor are all posts about Bush or the Web properly tagged as such, so a classification is somewhat out of bounds. Ideally, though, the classifying approach can work in a semantic web context where there's a robot that compares one thing to another. This is what Last.FM, for example, does with musical tastes based on the profiles its users build. If Last.FM did that with the tags, then, churning them through a similarity engine, you'd begin to have something at a low-level of semantic value. This is where Flickr is going.

Similarly, Web 2.0 -- which no doubt originated as a marketing term -- is not well-defined and little more than a "label" rather than a "classification" (here's where I go meta myself). The arguments and snark usually come from a classificationist perspective, but it's quite interesting -- in a meta sense -- how the originators of the term came up with it: they listed things that were "Web 1.0" and things that were "Web 2.0" and used that list to create the differentiation. Humans seeing the list can instantly grok what they're getting at, but it's lousy from a classificationist standpoint, as the tussles over that Wikipedia article suggest.

So what's funny is that they used a semantic framework to define "Web 2.0", which is a concept built on the synergies you can get from folksonomies, which are constructed in the same way that the definition of Web 2.0 was. I find that intriguing, if not -- ahem -- definitive.
posted by dhartung at 1:52 PM on November 20, 2005


dhartung: I keep writing and rewriting responses, but I think the problem I'm having is that I'm not sure what you mean by classifying vs. labelling. A structured and controlled vocabulary doesn't dictate physical or virtual collocation, it merely structures the metadata assigned to a document. A tag is no more disconnected that an index term, it's just the the method of applying a controlled term is more constrained and, as a result, increases relevance and precision, albeit at the price of recall. And when I get more results for a search than I could read in a lifetime, I'm more than willing to sacrifice recall.
posted by stet at 5:30 PM on November 20, 2005


I think the thing that annoys me the most about "folksonomies" is the confusion of lay people (and professionals!) around what users are actually doing when they "tag" images at flickr. They are not, as the Cnet article suggests, "classifying".

Of course not. They're building the neural interconnections that will form part of the substrate for the first true artificial intelligence.
posted by stavrosthewonderchicken at 5:44 PM on November 20, 2005


« Older Precious Pet or Tax-Payers' Property?...  |  Herring break wind... Newer »


This thread has been archived and is closed to new comments