Screen cast on pushback and RDForms incl. a demo

Thanks to Keith I’ve finally found a nice capture tool (worth the 29$) for my Mac to do a screen cast on pushback and RDForms. Here is the result – enjoy ;)

Turning the read-only Web of Data into a read-write Web of Data (the demo part starts around 3min 30sec):

We introduce pushback, a method that enables writing changes to non-RDF sources such as flickr, Twitter, Amazon, etc. from an RDF document. The video explains our motivation, the architecture and the interaction between the components as well as RDForms. A demo (for Jira, a professional issue tracker system) is included in this video, where we show how to create, deploy and use an RDForm.

As a side note: the last 5sec or so are somehow crap; I was hoping vimeo would allow me to trim it, but …

How does pushback relate to …

From the day we initiated pushback – on that day the idea has been discussed the first time properly in detail with Richard here at DERI and with TimBL via IRC – people would challenge me to explain how pushback would fit into the current schema. We can use proprietary APIs such as the flickr API to change or insert data in Web 2.0 data silos on the one hand and we have SPARQL Update on the other hand to update the RDF data in, say, a triple store (for example, ARC2).

But what if we’re surfing in the Web of Data, that is, viewing an RDF document using for example Tabulator? What we’re actually able to do nowadays is to update a native RDF data source (with SPARQL Update). But how many sites or services in the real world actually use triple store? You get the point …

On the other hand many Web 2.0 sites provide HTML forms for the ‘updateable’ part of their data; this could be an order form at Amazon or a Twitter post. In the linked data world people have invested time to create incredible useful so called linked datasets. One can understand some of the dataset as sort of read-wrapper around Web 2.0 data sources (e.g., Alexandre’s flickr wrapper). In the same way we’re able to set up write-wrapper that know how to handle the according Web 2.0 data source. We then just need a flexible and generic method to talk to these write-wrapper based on the data at hand, which means that the starting point in the context of pushback is always an RDF document (virtual, i.e. via GRDDL or SPARQL DESCRIBE or whatever doesn’t matter to us).

Take everything from above, shake it swiftly and voila, there you go (this is a morphological analysis, introduced by Fritz Zwicky a couple of years ago) :