Linked Data in Kasabi

Posted on 08/12/2011 by


There have been some questions recently about Linked Data support in Kasabi, so in this post I wanted to provide a few pointers on how we support Linked Data publishing, and how to access the data.

First of all, I’ll note that there’s a screencast that shows a complete worked example of how to publish Linked Data in Kasabi. You can find this on the TeamKasabi YouTube channel, and I’ve embedded it below:

Secondly, there is some written documentation on Linked Data Hosting in Kasabi. That document provides some additional guidance and notes on how to set up your data such that Kasabi can handle all of the URI de-referencing, i.e. so that users can follow links to your data via an application or in their browser.

While we use API keys to manage access to the APIs provided by Kasabi, e.g. Search and SPARQL endpoints, we don’t require API keys when accessing Linked Data. Anyone can access that data very easily.

In future, for commercial datasets, data publishers may require authentication to Linked Data, but that the default will be to support basic URI de-referencing.

At the moment to have Kasabi serve all of your data, you need to base your URIs on our domain, using the following pattern (check the docs for more information):[dataset-id]/[user-defined-path]

In future we intend to support custom domain hosting. This will allow data owners to maintain full control over their URI space. This will be a chargeable feature and will work similarly to a WordPress hosted blog. But right now Kasabi provides the simplest, easiest and free way to publish 5-star Linked Data.

Of course, it’s also possible to use your maintain your own URIs and use Kasabi as a storage layer, e.g. using the SPARQL endpoint or Lookup API to fetch the data to respond to requests. This will need some configuration on your server to manage the HTTP requests. Feel free to drop an email to the developer list if you want some advice on how to set that up.

One of the aspects of the Linked Data support in Kasabi that I’m most pleased about is that the entire dataset directory is also available as Linked Data.

If you visit then you can find a list — available as RDF via content negotiation — of all datasets in the system (94 at the time of writing).

That list provides links directly to the machine-readable metadata for every dataset. We’re exposing this metadata using the VoiD vocabulary¬†which lets us describe the dataset; provide links to all of the available APIs; and links to example resources provided by the data publisher.

We recently expanded the automatically generated VoiD descriptions to include some basic statistics including entity counts and the occurences of different RDF types in the dataset. By way of an example, here’s the description of the Yahoo Geoplanet dataset.

All of the above metadata: the directory, dataset descriptions, and Linked Data is available without an API key, so you’re free to explore using any existing Linked Data tools and browsers. For example, I’ve recently been harvesting our own dataset metadata to explore some ideas around dataset summaries and recommendation.

Want more flexibility to query and explore a dataset? Then sign-up to get an API key to access our standard APIs or custom APIs created by the user community.

Hopefully that helps clarify how Kasabi supports Linked Data publishing! Happy browsing.

Posted in: Technology