April 13, 2009
Does Everything = Data?
kidehen posted a tweet related to #linkeddata that I want to examine, but I can't do it in the 140 character Twitter limit. He says:
"Ground zero for me: data is everything, and everything is data. To make data useful each datum needs to have Identity."
I disagree with the first sentence because boiling everything down to data seems to conflate two categorically different kinds of things: Real World Objects and Web Documents.
I basically agree with the second sentence. The reason is that we are looking at two sides of the same coin: the thing itself (the RWO), and information about the thing (the datum). One side can't sensibly exist without the other. This datum can be represented in a variety of ways, which is the tie-in to Web Documents and the "old web". With this in mind, I might rephrase the 2nd sentence like so:
"To make a Real World Object's identity reusable it, and thus its datum, needs to have an actionable identifier on the Web."
5 comments so far
A Web Document is data i.e., a container of items of data. It has addressable location on the Web exposed via its URL variant of URI. But that doesn't imply that is can't have a URN variant of URI constructed using HTTP which is basically what happens re. Real World Objects in the Linked Data realm.
Maybe I should say: Everything has a URN variant of URI when it comes to Linked Data :-)
Kingsley
Kingsley,
I quibbled about your statement "data is everything, and everything is data", but I suddenly recall Andy Houghton and I saying something eerily similar in the past: "Everything is a resource". I still believe this is true and that there is a subtle difference between a "resource" and "data". Whether the blog entry above clarifies this or not is another question.
Jeff
Kingsley Idehen says: "A Web Document is data...", actually a Web Document is a representation of a datum which is comprised of a collection of properties. The Web Document itself is not the datum, but merely a representation of the collection of properties.
One can think of Linked Data in these terms: 1) A Real World Object is an abstract class in an imperative programing language, 2) A Generic Document is an implementation of the abstract Real World Object which contains the datum, 3) Web Documents are sub-classes of the Generic Document which produce representations. Since all Web Document inherit from the Generic Document they all share the datum.
Kingsley's statement: "But that doesn't imply that is can't have a URN variant of URI...", seems a little befuddled, but I think what he is getting at is that a Real World Object URI can be any URI scheme, which is true. However, if the URI scheme is not http, then you have no mechanism to connect the Real World Object URI to a Generic Document URI or a Web Document URI. Resolution is a fundamental aspect of Linked Data and is why Tim Burners-Lee four Linked Data principles say to use http URIs.
If we accept a "datum" as unit of observation, then giving said datum a name makes it useful (e.g. you can reference it). Resources are useful items of data.
When you use HTTP to construct the name of a piece of data, you then extend the range of reference to the Web (at the very least).
The reference to URN is about naming,
Linked Data (as espoused by TimBL) is about HTTP based URNs. Put differently using URLs to deliver URN function. Which is why 303 heuristics have to come into play when enforcing the notion of datum identity.
Remember, URLs and URNs are URI variants :-)
Kingsley
Kingsley,
Your use of the phrase "HTTP based URNs" makes me realize that we are out of sync in our terminology. It appears you are using the term "URN" in the classical sense of name vs. location. Andy and I use it in the contemporary sense of referring specifically to the URI scheme "urn:".
Resyncing our use of these terms would help in the #linkeddata discussion.
There also seems to be a subtle difference in our understanding of "datum". Reconciling this probably deserves its own blog entry, though.
BTW, thanks for your feedback! Seeing this from different perspectives is a huge help.
Jeff