1) What is the difference between Yahoo Pipes and DERI Pipes?Yahoo Web Pipes are great to operate on source which can be easily reducted to the RSS paradigm (item list). This does not map well to graph based data models such as RDF (RDFa), complex Microformats etc. For this purpose DERI Pipes have been written from the start to operate also on complex data, e.g. offering specialized operators which use RDF as a ceneral model to perform data aggregation and transformation tasks. 2) What is DERI Pipes Licence?DERI Pipes frontend and GUI editor is licenced under the GPL licence. DERI Pipes runtime is distributed under the MIT licence. 3) Can i create Pipes without the frontend?Yes, DERI Pipes runtime runs pipes defined in Pipe XML language. It is easy to get the XML from the GUI editor 4) How large can the sources be? Please don't try to load large sources on this public pipe server. If you want to use pipes for very large sources, make sure you get the latest pipe engine distribution and run it on your local resource. Usually sources up to 100k are ok. 5) I got this great operator idea Sure, let us know or join the effort! Or a simple way to do it is to define your own pipe called "MyOperator" and to make this fetch teh result from your operator implemented online on your server. Soon you'll be able to wrap that into a pipe and use it as a block inside other pipes.
6) Other Related Works? The term "Pipelines" or "Pipes" in Computer Science is famous for its used within UNIX environments, but is a well known general paradigm.
Yahoo Web Pipes is also very relevant as it proposes a similar paradigm, however with operators which are in general meant only to process RSS feeds. The need for a cascade of operators to process RDF repositories is also addressed in the SIMILE Banach project. Banach operates inside the Sesame DB by leveraging its capabilities to have a pipelined stack of operators (SAILS) which can both process data and rewrite queries. Similarly, cascaded XML transformations are sometimes referred to as XML Pipelines. There are other projects out there, we're sure. Pleasesend us links and we'll put them here. 7) {Semantic Web Geeks}Why are SW Pipes not written in RDF? RDF is a very smart choice for knowledge representation. Wisdom says, however, that RDF should not be used if you're not ready to do a proper conceptualisation of the entity relationship model behind what you want to represent (e.g. creating a proper ontology). Such task is often all but trivial. Furthermore, given RDF simple basic nature (triples), a proper conceptualisation of something as inherently complex as a language will often necessarily yield to very complex representation structures (e.g. involving chains of blank nodes), which are not at all handy to work with. Giving up on such proper conceptualisation and just going for idiosyncratic semantic constructs in RDF would simply be useless as the result would very likely not be usable or processable with standard tools like reasoners, query languages or basic RDF operations such as a merge. Preaching aside, if you think a GRDDL transformation could be of use, let us know and we'll see what we can do ;-). 8) {RDF Geeks only} Do you offer a SPARQL endpoint for each pipe end? Not yet, but that's coming next.
|
