Slashdot Mirror
How Google Will Have Achieved The Semantic Web
alfaromeo points to a business feature (mysteriously available already) by one Paul Ford called "August 2009: How Google beat Amazon and Ebay to the Semantic Web." So read on for a bit of potential history from five years in the future.
Remember back when we all thought that XML was going to achieve the semantic web by making good search engines unnecessary? Now XML has gone nowhere except as a set of popular libraries for cross-language data serialization, and we're starting to talk about just making really smart search engines.
Very interesting ideas, but I seriously doubt that Google could (or would) try to squeeze a percentage out of every transaction performed using the hypothesized marketplace manager. That just doesn't seem to fit their modus operandi. More likely they'd give place preference to paying clients, much as they do now with the existing search pages.
But as I said, a provocative read. Metadata truly is the future.
Interesting prediction there ... but what does it have to do with The Semantic Web? Oh well - guess it's pretty hard to write a fictional future piece without injecting bizzare humor into it. Right? Right?
I hear there's rumors on the Slashdots
didn't so much refer to XML the technology as to one of XML's proposed applications. There was a popular theory within the press when XHTML was first introduced that XML would supplant webpages and drag the web back to that primordial point when HTML was intended as a content markup language, not a display language, and even go beyond that. Supposedly we were going to wind up where stylesheets would go beyond just a mapping from XML tags to some set of HTML4 tags, and into a point where content was just a minimal set of XML-tagged text and everything about the way the site displayed was deferred to CSS-like technologies. And when this happened supposedly web browsers would be totally free to reset stuff, and we could toss out amazon.com's presentation of, say, the search results for "Michael Jackson" (as a series of paragraph-delimited links to categories (books, music, etc) to search within in a blocked-off area surrounded by amazon.com's navbars and logos, which then pointed to a series of pages containing little formatted blips of information about various items for sale presented in groups of ten separated by little gray lines in a blocked-off area surrounded by amazon.com's navbars and logos), and instead have it display as a heirarchial file browser or whatever we liked.
Well, I think it's safe to say that idea's been mostly shelved for the time being. This isn't a matter of a lack of "reaching potential", it's a matter of total failure to move in that direction. XML has been incredibly popular as a storage mechanism but has had roughly zero takeup as a communication mechanism. (There have been communication substrates, such as XML-RPC, based off of XML, but that's not the same thing.) I don't know if it's fair to assume a technology come to fruition within 8 years of being proposed, but I think it's fair to assume that unless we see some kind of signs of progress or interest in progress within 8 years, there's no reason to expect further progress within the 8 years after that.
A babelfish English->Russian->English translation works out as "Flesh is willingly ready but spirit it is weak", which is pretty close to the mark.
SCO employee? Check out the bounty
No it's not. Maybe you're confused with VVV which is 777 in Arabic figures. Numbers in Judaism are either written in Hebrew letters, which would be (for 666) Taf-Resh-Samekh-Vav. If you really want to try and pronounce that, it's read Tarsav.
Good luck finding a conspiracy theory in that.
And oh, 666 is a christian number.
"Programming is life, the rest is mere details"
Strong AI requires grounding symbols in real world things, events, and processes.
I think that simply defining the "meaning" of words in ontologies is likely good enough for useful web-based software agents. It will take time, but with well defined ontologies, and common use of RDF using standard schemas will make a lot of cool things possible. I think that dealing with ungrounded symbols, but symbols defined and related to other symbols in a structured way, is OK.
One of the classic complaints of AI systems can be summed up with a trivial example:
Define a relation in Prolog:
father(ken, mark).
A human reader assigns their own meaning to "father", "ken", and "mark". To a prolog system, this could just as easily be:
aaa1(aaa2, aaa3).
Somewhere, on the edge of symbol-slamming systems, there has to be some connection with the real world, with our experiences.
For semantic web applications, this "edge connection" can simply be tying into symbols defined in OWL ontologies, RDF Schema, etc.
The problem is getting people to use RDF (I added RDF to my main web site years ago, but it only contains limited information).
Another problem with RDF is that there are several kluges to get it into XHTML, but that will hopefully change soon.
A good toolkit for experimenting with the semantic web is the Swi-Prolog semweb library (http://www.swi-prolog.org/packages/semweb.html/)
-Mark
And oh, 666 is a christian number.
True. I watched an interesting television program that suggested the number 666 was part of a game people of that era used to play. They'd take a person's name and sum up the values of the characters in that name. The program suggested that Nero = 666. So by saying "666" Christian's were making a somewhat encrypted political statement against the Romans. It continued to say that because of a mistranslation, apparently the number is actually 616, which is the sum of another un-popular Roman emperor, Caligula. I don't remember the actual math so don't take this as the final word on the matter.
A system that perpetually collects information presented in a language that easily conveys the attributes and logical relationships between different objects and concepts. (Scratches beard.)
... !! ... -- THERE IS ANOTHER SYSTEM --
Make the system distributed and let people run their own information collecting agents. Every home computer becomes a part of the network of logical relationships, each with a tiny piece to contribute to the puzzle. My computer could have complete information about the workings of combustion engines - what parts they consist of, and their relationships.
When someone requests information about car manufacturing, some relevant part of it will be retrieved from my store.
Now, let's make the system ask us for help, when information is missing. Let the system start drawing own conclusions from the facts it gathered, and tell us when it needs something filled in. As it grows, more and more complex queries could be answered.
Q: CAN THE EFFECTS OF GLOBAL WARMING BE REVERSED?
A: THERE IS YET INSUFFICIENT DATA FOR A MEANINGFUL ANSWER
Or how about:
A: TO REDUCE GLOBAL WARMING, FIRST WE MUST... ??
Oh, at least I hope the network will be able to finally find the true correlation between the price of gold and the length of men's beards.
Reasons:
- It relies on said dreamweaver jockeys bothering to do this at all, let alone correctly.
The real semantic web will involve AI spidering and parsing of human-readable web pages. It will be as inaccurate, but as useful as babelfish. It's the only answer that makes sense -- because that's where all the juicy data is.I was actually a bit disappointed by the article. First of all: it is very hard to search in distributed knowledge networks, if not impossible. Some structures, which are a necessity to make explainable in an onthology are possible to describe, but not possible to make deductions on (some of the queries cannot be proved to finish at all). An example are meta-classes (a Chardonai wine can be an instance of the class Wines, in which case a specific bottle of wine can be an instance of Chardonai as well as a normal wine).
Second of all, the article fails to mention anything about the Ontology Web Language (OWL, see this site on W3), which has become an official specificion of W3C since May this year. This language, based on RDF is much more expressive than RDF is, it also contains several 'language levels' based on the amount of complexity and decidability involved.
Last, but not least: the article is still very vague on privacy and thrustworthyness. I would think that public-private key cryptography would not do in these areas: far too many single points of failures when, for example registering. Only one user with a hacked account can derail the whole system!
I'm really interested, by the way, to speak with some people who are deep (at least above their knees) in OWL and RDF. Planning on making a study at intelligent databases and datamining.
This is a replacement signature.
Microsoft Bob succeeded, but not in the way you have expected.
Melinda Gates (nee' French) was the Product Manager of Microsoft Bob.
(just don't brag to your friends you've known that forever)
p.s. Microsoft Bob is|was one of the products (along with things such as RedHat) which Virtual PC can run successfully; so it hasn't disappeared completely. I still have a copy sitting here in one of my CD wallets. (Handed out at a Tech Ed or some other conference)
Gematria, or the use of the numeric values of Hebrew letters as a means of interpreting the Torah, began being used around the Greco-Syrian occupation of Judea and Israel. The term is actually a corruption of the Greek term "Geometria" or the science of Geometry. The transliteration of Nero Augustus Caesar into Hebrew letters can be done in two ways: one gives the value of 666, the other 616. I don't know if Caligula Augustus Caesar works in the same way but Nero works both ways.
The book "The Revelation of Jesus Christ to John The Divine" is a very interesting one if looked at not as prophecy (anyone and anything can be worked into the fantastic account) but as historical allegory. It seems to have been written to encourage the persecuted Pauline Christians, who were persecuted with great vigor around the time the book is said to have been written. Just google on Nero and 666 and you can read some very interesting stuff.
Knowledge is power. Knowledge shared is power multiplied.
Yeah, but your scientists were so preoccupied with whether or not they could, they didn't stop to think if they should. -- Dr. Ian Malcolm (Jurassic Park)
This would deliver the invistigative powers of the CIA into the hands of anyone who wants it... still a good idea?
No kidding! conspiracy theorists are already having a field day And senators are alredy trying to lift the restriction on foreign-born people becoming prez.
The article failed to mention flying cars, another no-duh prediction that seemed completely obvious, and won't happen either.
A short while ago, Cory Doctorow published an piece entitled Metacrap: Putting the torch to seven straw-men of the meta-utopia, which mentioned two very good reasons why the semantic web won't take off the same way that these articles predict: schemas aren't neutral, and there's more than one way to describe something. These are basic problems that have been hounding AI research for years, dictionary & encyclopedia publishers for centuries, and all other academics for millenia, and they aren't going to go away.
The central problem with universal metadata is that it requires informed work on the part of data creators, and it's a major pain in the ass. The OED took almost a century to create, and the first few decades were essentially wasted figuring out that dilletantes were not adequately capable of properly cataloging use of language. Even with a profit motive, good metadata is a bitch (see EBay comment in the article above).
It's like the senator's (I forget who) comment about pornography: "I can't define it, but I know it when I see it." Often, we don't know what it is we're looking for exactly, and we don't know how to describe what we've got so other people can find it except in very narrow terms. I have a few creative projects which I've released under the creative commons license and dutifully marked up with cc's provided RDF information, but all that code just says what the license is, not what the project is like in a way that's as meaningful as, for example, a music recommendation from a friend who knows your tastes. The porn industry (as usual, on the bleeding edge of information and communications technology) deals with this to some degree by having a very narrow semantic universe to describe: Search Extreme is a stupendously complete metadata set, but even it contains only a few kinds of information.
Don't say that to real semantic web junkies...RDF is most decidedly not XML. There is a format for serializing RDF called RDF/XML, and that is indeed a common way of passing RDF around, however RDF really is a number of statements, each with a subject, predicate, and object (like a sentence).
XML is more of a key, value type of thing, and as such, without a priori knowledge of the meaning of the key, a computer can't reason about what the value means.
All joking aside about the above comment, it raises a question about "Semantics" in general that NO meta -data system no matter how good you make it is going to address...
If Jim has a friend, your talking about an expression FROM the perspective of Jim. This expression is is given CONTEXT by Jim, it can NOT be said that Paul has a friend named Jim... Paul might not know Jim (stalker). I could poke holes in this for days but I dont't rreally have the time.
Why cant the W3C come out with some standards that are USEFUL. How about some tags for address, phone umber, geocode, so I can search on what is "local". How about some working samples that are publicly available, so people can start putting them to use.
-- RANT FOLLOWS --
I'm beginning to view t.b. lee like I view jacob nelson and Steve balmier, just another talking head throwing anything and everything out there and hoping it sticks. (picture all 3 of them on stage, Steve screaming "developers" Jacob saying "usability" and Tim going "XML")
It seems that Tim has forgotten how he got into the position that he is in, by putting something out there that WORKED (html) and letting it evolve from there. We had to live it for a while, let others take it up before we could get to a better place. It is starting to look like the W3C has over stepped it posisiton as the shepard of the web to try to be an innovator, and I don't think that is the ROLE a standards body should be assuming. I would really LIKE to see them (the w3c) stop wasting money on crap that 90% of the world is NEVER going to use, and start trying to make things more accessible to Joe average user. Most of the "standards are good and you should use them" has come from small iconic developers (zeldman anyone), while we get OWL from the w3c? How about making some of those wonderfull standards clear and accessible, how about a solid explanation of semantics for Joe average developer, who could not read through the "doccumentation" to save his life....
-- end rant ---
Funny you mention an Isaac Asimov character. I remember a short story of his called "Sis" about an orbiting computerized that took over the world in a benevolent sort of way. IIRC, he ended up comparing it to God as all-knowing, all-powerful and all-good. If Google sticks to their "Do no evil" policy, maybe they will become "Sis."
While that article is interesting and all, the author is pretty quick to say how Amazon didn't embrace the semantic web.
Amazon is the best (most useful) application of the theory and technology behind the semantic web that you will find anywhere right now. Granted, I don't *know* exactly how they are doing what they do, and its not a "public" interface in the way that the semantic web is envisioned, but it is a large scale implementation of knowledge management principles.
Did you ever notice that whenever you look at a book (or anything really) on Amazon, it gives you suggestions for similar books, suggestions for books that other people looked at who also bought that book, suggestions for books on topics that you have previously bought books for, etc? The semantic web is at heart a directed graph. Amazon is at heart a directed graph, too. Their graph grows every day with new knowledge based on the actions of people shopping on Amazon, and new conjectures about the relationships between products can be made by simply walking that graph, and computing the transitive closures of the statements (ie John likes the things that Mary likes, and Mary likes Jane's taste in music, so John may like the music that Jane bought).
This technology has incredible power, the ability for a machine to draw conclusions like that. Do I think that it will work the way that article thinks it will? No, not if the masses are left in charge of the metadata. It works very well for Amazon because they can control the quality of the metadata, so erroneous conjectures are not made on bad information. I don't think Google is by any means _not_ paying attention to the semantic web, but I think that Amazon is already there and has been for quite some time.