REST/Linked Data

by Saurabh Gupta

It wasn’t hard to think about RESTful concepts when you hear Tim talking about linked data…

and a quick search and bang .. people are already talking about it here and here. The first link is exactly in lines of what I thought – huh isn’t he talking about REST? . Then the second link attempts a nose dive into conceptual paradigm.

Any case, as I understand it and may be I can explain to others is this. Any data driven web-page, in an email service, or in a wiki or in an imdb… is representing current state of data. This representation is probably also uniquely identified on web. So if I have to go to Kill Bill Vol 2 again on IMDB i will use this instead of searching it within IMDB again. This style of architecture is called REST. This representation of data can take multiple forms. It can also take multiple levels of granularity.

Multiple levels of granularity! This is where Tim Berners Lee wants to extract the power of web. Every conceptual information represented by a URI (unique resource identifier) on web. This conceptual piece of information, if requires to, is linked to another resource anywhere on web using another URL. The concept is no different  than current state of affairs with hypertext linked web pages/blogs/twitter etc., but one key difference. Instead of linking set of web-pages/documents/tweets/blog-posts, we are linking together pieces of information or raw data.

This is powerful. So instead of linking up bunch of documents together for the reader or a bot (like google, yahoo, etc) to understand the overall context, this way we are building up the train of thoughts or ideas or logic with narrower margin of possible deviation in interpretation.

Concluding with what Tim calls “Raw Data Now”. It is not uncommon among everyone of us to dress up before making a first contact. The websites are dressed up to look nicer. Representation of data follows later. However, representation of data – how much ever raw the format is – should be up there before you have figured out a way to present it in 5 stars. This is particularly important for public enterprise, government, empirical data, scientific journals etc.