Monrai Blog

News about Cypher, Semantic Web, Natural Language Processing, and Computational Linguistics

Sunday, June 07, 2009

Razorbase Safari Bug

UPDATE: The Safari bug is now fixed.

There is currently a bug in the xslt stylesheet for Razorbase which prevents Safari from displaying the main results page. No resolution is availbile yet, but I'm working on it. I apologize for this inconvenience to Mac/Safari users. I'll post once the bug is fixed.

Labels: , ,

Saturday, June 06, 2009

Razorbase Examples Part 5

If the litmus test for a killer app is the ability for users to create fun, ad-hoc games with it, then linked data is the Semantic Web's killer app. Here's an example of a variation of the Kevin Beacon game using razorbase.

If you discover any new games using razorbase, please post the rules as a comment.

Labels: , , , , ,

Friday, June 05, 2009

Razorbase Examples Part 4

In these slides, I walkthrough the use the file system metaphor in the razorbase UI.

Labels: , , , , ,

Razorbase Examples Part 3.5

I've added "type actions" which allow you to do multiple-step walks through the database with one click. Currently, actions have been added for types: Category and Person. More will follow soon.

As a side announcement, a re-crawl of the LOD is expected to be complete between today and mid-next week. This means less errors in the data, more current data, and less mal-formed characters.

Thursday, June 04, 2009

Razorbase Examples Part 3

There is a ReadWriteWeb article about Google Squared, a new labs product that gives structured results to keyword search. It's Google's version of razorbase. The article expresses some of the services shortcomings using a search for dog breeds as an example. In response, I've created some slides that shows how a lookup for dog breeds works in razorbase. It's a great example of the universe of Linked Data vs. the closed world of Google's document web.

Labels: , , , ,

Tuesday, June 02, 2009

Razorbase Examples Part 2

I've released a second presentation on razorbase, this time showing the alternative ids action in action. I also give a demonstration of what isn't allowed to happen when people don't adhere to the Principles of Linked Data.

The world is refining how it consumes, people are demanding goods and services to be delivered in ever decreasing sizes that allow the consumer tighter control and choice over exactly what is consumed and how much of it. Cell phone billing went from long-term service contracts that spanned years, to pay-as-you-go plans and the concept of more grainular roll-over minutes gained popularity. Amazon has figured out how to break the whopping utility bills of their server centers into small, per-usage units which the consumer pays for as they go. Youtube turned the world of media on its head by allowing users to instantly share bit-sized video clips that can be consumed faster than it takes to clean your email inbox. The notion of buying a record album has been usurped by a iTunes feature that allows for purchases at the level of a single track of the user's choice. One could refer to this trend as the atomization of goods and services, and it's about to happen in the realm of the WWW (and data in general), and in a very big way.

I've created and used many Linked data/Semantic Web demos over the last decade, and each has taken a level of... imagination, in order to feel where this is going. Razorbase is the first time I've been able to really sit back and see our destination physically on my computer screen. The single most revolutionary feature of this service is the Information About page, which lists all properties of the subject in focus. If the subject contains a list of things, then you'll see all the properties belonging to all the results. Most linked data browsers I've seen attempt to push the entire description of a URI resource into the UI at once (or at least as much as one results page can hold). This is a very pre-atomization approach, akin to how a Wikipedia article is displayed. If you look up WWII on Wikipedia, you could easily spend an hour scanning the article and taking notes. The user would benefit more by having this information sliced and diced into bit-sized, labeled chucks, and placed on a platter for them to choose freely from... in a word, atomized. As long as the user has a general notion of what properties are (probably) availible, and what they can expect to be on the other end of those properties, then the user is able to (in most cases) easily build a mental map or strategy of how to get to the exact information they're interested in, no matter how complex or esoteric the connections involved, no matter how heterogenous the sets of data are standing between the user's start and end point. Once the user has a mental map, they only need 1) paved roads and 2) the right sign posts to keep them on their track. Razorbase provides sign posts by displaying blue arrows that take the user back and forth through the dataspace. It provides paved roads by e.g. using collaborative-filtering-esque techniques for exposing the most worn paths through the dataspace.

Labels: , , ,