Slashdot Mirror
Google Acquires Metaweb
eldavojohn writes "A startup called Metaweb (looks like an ontological, entity-based approach to Web 2.0 tagging) has been acquired by Google. You can find out what they're about from a super marketing fluff video they put together. The neat thing about Metaweb is that the database of entities it has is free. Will Google be able to make Metaweb work on their omniscient scale, or was this just Google making sure a startup doesn't become yet another player in search?"
Will Google be able to make Metaweb work on their omniscient scale, or was this just Google making sure a startup doesn't become yet another player in search?"
If Metaweb doesn't work at Google's Scale, then it couldn't compete with them.
Everyone was thinking Google would take over the Web, and here they skip right past it and acquire the Metaweb.
Well played, Google, well played.
How can I believe you when you tell me what I don't want to hear?
i didn't' like.
L'esperienza de questa dolce vita (The experience of this sweet life) - Dante Alighieri, The Divine Comedy
I always thought "Ontological" was a synonym for a useless philosophy degree.
I judt got a nre Kinesis keybiartf so please excusr ant egregiou typos.
Will Google be able to make Metaweb work on their omniscient scale, or was this just Google making sure a startup doesn't become yet another player in search?
Wrong and wrong, you see Google is freebasing now:
The web isn’t merely words[, or water-soluble,] it’s information about things in the real world, and understanding the relationships between real-world entities...
Sometimes you have to give it a good ole "smoke-test" to see the possibilities...Google should be careful though, the path they have chosen is a slippery slope!
We should start a new Slashdot and return control to the geeks. It actually wouldn't be that hard to get some users to
They sure have an ugly web page.
Looks like this may be a way to make a play for competition in homeland security and business support, like Palantir has done plus medical data tracking, and other possible extrapolations
I'm fairly sure it's not going to be used for just generating websites.
First winter rain-
even the monkey
seems to want a raincoat.
-Basho
Slowly but surely google continues to acquire startups and expand their business. Not that I mind it that much in Google's case but isn't this the type of thing that Microsoft or AT&T eventually got hammered for?
Legitimately wondering if Microsoft and AT&T did it much more dastardly or if there's no significant comparison whatsoever.
One of the challenges with generating and using data sets is cleaning them up. Data entry errors, OCR failures, conflicts between multiple sources, etc. make it a pain to search and summarize data. Gridworks helps me hunt down bad records and normalize fields. If it keeps improving, people might start using it before publishing their crap data.
I Browse at +4 Flamebait
Open Source Sysadmin
The problem is that nobody wants to express information through RDF tuples and ontologies. Instead, they express information in human-readable text, with structural and visual markup. Search technology has come very far in terms of figuring out what information we actually want, with things like personalization, disambiguation (see DuckDuckGo for example), shopping/product search, and so on. All this stuff can be teased out of traditional web content with far less effort than trying to get every company and individual to express information through formal ontologies, etc.
So yeah, basically, the goals and use cases of the semantic web fall into two broad categories 1) stuff that data mining or search can do now and 2) stuff that requires hard AI or tons of human labor and thus won't be happening any time soon. This is why "ontologies" have become synonymous with fail.
"Ontological" is an essential adjective for describing different aspects of knowledge (science); ontology for ordering it.
Intelligent idiots are we. | Evil men do not understand justice.
stuff that requires hard AI or tons of human labor and thus won't be happening any time soon.
Wikipedia.
Was anyone else amused by this? (RTFA)
"Going to war without the French is like going deer hunting without your accordion." ~General Norman Schwarzkopf
In a way, I miss Alta Vista, in that they had a few things that Google does not:
Say you searched for "wine", and activated that mode. It would present you with some possible extra terms you could search on, such as "white", "red", "tannic", "windows", "microsoft", "emulator".
Were you to be searching for the fermented beverage, you could select "red", "white", "tannic" and so on.
Were you searching for the ABI adapter package, you could select "windows", "Microsoft", and "emulator" (which yes, Wine is NOT...)
I'd love to see Google add that sort of refinement, ideally "learning" what sorts of terms go with what (Wine + tannic = beverage, wine + OLE = software).
www.eFax.com are spammers
I was on the founding team at Metaweb when we spun out of Applied Minds. I can answer some questions here, but first I wanted to congratulate the team that brought this company all the way to acquisition.
So, from the beginning we knew that semantic this and ontology that would be a non-starter for most contributors from Planet Earth. While Freebase is a complex system under the hood, the user interface makes contributing data to an existing type (schema) pretty easy. You can add content from a browser window and never know that all of your entries are typed by the system. You can upload a spreadsheet of data and not have to do anything more than say which column is linked to what field in Freebase.
My startup, 24 Hr. Diner, uses Freebase to demo our artist to artist recommendation engine, Jukebox. We have recommendations for 100k artists, and for each of them, we can look up their genre info and photo on Freebase without having to maintain all of that data ourselves.
And if anyone on Slashdot is working for a co. that could use an excellent recommendation engine that handles music, videos, and general web content, ping me!
Well, a thousand years ago computer science would also have been a synonym for a useless philosophy degree.
This is why "ontologies" have become synonymous with fail.
So you're saying that Google bought a failure to save the rest of the world from it? It's the "tons of human labor" part that becomes the issue; it's bad enough trying to teach a human about semantics, let alone a pedantic automaton. Wake me up when an AI can disambiguate without me spending 45 minutes explaining the basics of English language.
"Please describe the scientific nature of the 'whammy'" - Agent Scully
Early on, we knew we'd have to make a UI so that users could have as close to a free-text experience as possible while still contributing structured data. Freebase lets you create a topic that is generic, and then co-type it with multiple specific types later. It allows ontology geeks to do their thing, and regular users to just work where they are comfortable. It's a tough balance to strike, but Metaweb's Freebase was populated by a small team of data wranglers using a mix of automated methods and coordinated manual cleanup and entry, along with power users who were especially interested in particular data domains.
At one point I was really interested in submarines. I created a type describing the key characterists of subs and then spent a few days finding all the generic topics in Freebase on subs (many from Wikipedia) and filling them in. Others, either at Metaweb or outside, have done similar efforts on other domains.
Few contributors ever say or even have to consider ontologies. If they want to dig in, it's there, but almost never presented in a way that requires a PhD and a pipe.
Look forward to Freebasing with Google!
my sig
All your (free)base are belong to us.
... which is exactly what DuckDuckGo uses as its data source to handle disambiguation. But Wikipedia is structured for humans and features a large volume of knowledge in human language form with some basic markup. It's not a bunch of information encoded in RDF tuples. Thus my point. Trying to get everybody on the web to re-encode the vast body of knowledge out there in RDF, explicitly referencing ontologies is a setup for failure. Sure, you might use some sort of tuple format to internally store information that you parse out of the human-language web, but that's different from what the "semantic web" set out to be initially.
Aaahhh Wikipedia... the idea of collecting “facts” by determining how many idiots did not disagree.
Or in other words: Argumentum ad populum hard at work.
Any sufficiently advanced intelligence is indistinguishable from stupidity.