Slashdot Mirror
Berners-Lee Pushes Linked Data In MIT Course
ErMKutz writes "WWW inventor Tim Berners-Lee is championing linked data — the idea of assigning web addresses to individual pieces of data to enable more intelligent information searches — much like he did now-ubiquitous Internet standards such as HTML and HTTP. But the ethic hasn't quite taken off yet, so he and a group of Boston tech and entrepreneurial all-stars are launching an MIT class to teach students linked data mechanics and fast-track the technology to market. They're combining engineering and entrepreneurial education in the hopes of launching viable linked data businesses or open source code at the conclusion of the course." I hope this shows up on OpenCourseWare.
Please bring back the BLINK tag.
So how do they connect
I am Slashdot. Are you Slashdot as well?
Chunks of data that are
I am Slashdot. Are you Slashdot as well?
apparently related?
I am Slashdot. Are you Slashdot as well?
Are HTML named anchors an example of data-naming? At least some browsers will render a resource around an anchor, if its name is given in the URL.
Applied to the web (and with a way to join two pieces of data) this can lead to a HTML-supported bottom-up approach, with no need for "a special way to #include files". People could then create welcome.html-piece, toc.html-piece, blogpost.html-piece and say index.html is *.html-piece.
This sounds pretty much like deep linking. which per wikipedia is
Deep linking, on the World Wide Web, is making a hyperlink that points to a specific page or image on a website, instead of that website's main or home page. Such links are called deep links.
I remember hearing about a couple of lawsuits which were raised because of deep linking and i dont see how this is any different.
I can faintly hear the lawyer sharpening their tools right now......
What the hell?! Is this something I'd have to read TFA to understand?!!
the preceding comment is my own and in no way reflects the opinion of the Joint Chiefs of Staff
The useful part is that anything that's natively on the Web is something which can be pulled into other completely unrelated applications, without having to wire them directly into whatever database you're running.
In other words, it seems a bit like REST or AJAX in that it's a new name for a useful idea which has been around for awhile, even in use for awhile, but needed a name and a slogan to become a concrete concept.
Don't thank God, thank a doctor!
One problem -- from a business perspective -- of linking data in a machine-understandable way is that it makes it much easier for third parties to use that data. At first that may seem like a good thing, but for many companies the data are the entire business. If a third party can quickly aggregate related data from many sites in a way that is more useful than the individual sites, those sources suffer. We're seeing this tension already with Google vs. publishers, where the data in question are news stories, and Google News is the aggregator. While linked data could be enormously beneficial for the user/consumer, the producers have little incentive to expose their intellectual property in a way that is easily used by other automated services.
Some of us what are old enough see so many "new" things that are repackaged "old" things that have been either forgotten about or simply over looked. Methinks this is another example. The implementation details may be different, but this idea was first promulgated in *1960*! http://www.xanadu.net/ refers...
There are inherent dangers of this level of linkage. One week,
person clicks link X: "We are at war with eurasia",
next week, clicks the same link "We were never at war with eurasia".
No one else see this? Archiving all information on the internet is one thing, but singularly cataloging and tagging every piece of information so that it can be accessed so easily is....well, dangerous.
While the irony of posting this on the internet is not lost on me, where all the collective information of mankind is at my fingertips, it just chills me to the bone to think that one day, phrases like "due process" and "freedom of speech" could be changed so easily.
"Linked Data is a sub-topic of the Semantic Web". Its latest technical effort, imho, is RDFa. So, its not like what you portrayed, more like a machine readable way to access web addressable data for re-purposing. See http://www.ted.com/talks/tim_berners_lee_on_the_next_web.html
The concept of linking huge amount of publicly accessible data is obviously worthwhile. The problem with the Linked Data movement is the current implementation. It is a total mess. The insistent attempts to pre-standardize open data have created a horrible bureaucratic monster. RDF, RDFS, RDFa, N3, RIF, SWRL, OWL, SPARQL, FOAF, SIOC, and a few others I forgot on top of XML. Every time you encounter a field with so many acronyms you know something horribly wrong is being developed. The consultants and enterprise "experts" will have a field day with this.
I wonder if there make the links rel="no-follow" :)
Office Photocopiers
The problem with most of the semantic web concepts is that it wants to add lots of metadata to everything, generally of a type that cannot be added automatically. I see how many people's music collections are extremely inconsistent with the metadata. Sometimes half the music is missing artist tags, and whatnot. Furter how many people actually have well tagged photo collections?
If people are not willing to tag their music and photos consistently, when it have active immediate benefits to them, what chance is there of getting them to tag information in their web pages, when that provides no immediate benefit, and quite possibly no benefit ever if the Semantic web does not catch on.
The idea of linked data on the other hand has a shot of working, but probably not in the way Tim Berners-Lee imagines it. RDF seems to have too much of a stigma attached to it.
Stylish sheet to fix many problems in Slashdot's D3: https://gist.github.com/801524
OK, I'll bite. What was B-L's 2nd successful contribution to the Web?
Changes are already made within OpenSim And Second Life that fetch assets not by file protocol and UD, but by HTTP. Opensim(second life free version) can even load entire region data via http URL(implemented) within the last month.
For those that dont know, 1/3 of Linden Labs have been made redundant, and some bloggers suspect this is in response to the sudden reliasation that its all about HTTP and not about propriertry viewers any more.
In the day I used to load TGAs via C. Its interesting noting the source code in Opensim that loads a textures via HTTP.
Almost certainly this is a response to the percieved advantage of being facebbok like(rightly or wrongly).
A quick and dirty example: A persons 3D avatar(Colada mesh file ) is stored on their website. As they visit a 3D world in browser, the world takes their metadata(loads the mesh via HTTP URL) and other participants see that representation.
In post Patriot Act America, the library books scan you.
Same as with Jesus Christ. After the New Testament, He did nothing worthwhile. I think it all went to His head.
As a college professor, I believe that the primary goal of a class should not be to advance your personal agenda. Feel free to share your opinions with your students, but your primary purpose is to inform and inspire, not to brainwash.
I'm clearly in the minority on this one.
This latest effort strikes me as a less ambitious version of the semantic web, and we may have seen a glimpse of it in Wolfram Alpha. Wolfram's ego aside Alpha does have merit and if the state of the art can advance as far as finding context in the written word then it's a damned good start; recognising faces and locales appears comparitively simple.
Call me a cynic if you want, but I strongly doubt that people in general will manually tag and classify their photos, movies, songs etc. This doubt seems ever more justified as the rate at which we accumulate personal data increases.
If God forks the Universe every time you roll a die, he'd better have a damned good memory.
There are a lot of companies and organizations trying to champion linked data, but linked data is nothing if those same companies and organizations don't adopt standards and push them ubiquitously. That was the motivation behind http://commontag.org/. It's a semantic data set of interrelated semantic concepts from various sources, but with a pretty impressive line of companies backing it up and implementing it.
Alpha is indeed a good idea, but I still lament the implementation.
Many times I've wanted data that can be trivially calculated from data it has (I can pull up each peice individually, so I know they are there, but there are a few to many to exract one by one and then calculate on them), but I cannot find any working syntax.
If I could have a programmatic interface to individual data points. My original thought was something like:
for x in ["USA, UK, JAPAN"]:
for y in range(1979,2009):
AlphaData("WolframAlpha.Countries.%s.government_debt.historical.%s",x,y)
However, exposing each piece of data with a unique URL that can be programmatically constructed would be just as useful. I would feel no need to have the returned data link to their locations, but I could imagine that for an example like this having some links like for the next year, previous year, next country, previous country, and perhaps several other links could be useful, and would probably be easy enough for them to expose.
As for tagging:
I agree with your doubt. Unfortunately machines are nowhere near good enough yet to do all the necessary tagging. Songs and movies can probably be handled by calculating good fingerprints, and looking up metadata in a large curated database, but photos tagging is much less easy. People usually want them tagged by both location, and people in the photos. The people in the photos is actually the easier problem, unless the photo source has GPS and geotags the image on creation. But even an algorithm to automatically identifies the individuals in each picture tagging them (prompting the user for a name if it lacks one for the individual) is difficult technology, and out current attempts are not particularly promising.
Stylish sheet to fix many problems in Slashdot's D3: https://gist.github.com/801524
Music metadata can be added automatically: see Picard and its music recognition system.
Dilbert RSS feed