Peter Ansell wrote: > On 9 April 2010 09:48, Kingsley Idehen <[email protected]> wrote: > >> Peter Ansell wrote: >> >>> Hi Mitko, >>> >>> Is this error by design in Virtuoso or is it just implemented for >>> DBpedia.org ? Seems a little arbitrary to stop serving information >>> just because it is more than 10MB, which is not that huge an RDF file >>> really. Compression will only make it slightly smaller, and in some >>> cases even the compressed information might be more than 10MB, how >>> does DBpedia plan to handle really large descriptions if they ever >>> come up? >>> >>> Cheers, >>> >>> Peter >>> >>> >> Peter, >> >> This is a setting on our part which is simply enforced via Virtuoso instance >> settings. >> >> As for descriptions, consumers are going to have to be smarter, there are >> many ways to get a description bar a mass dump of everything. >> >> 10mb is very generous bearing in my we are serving the whole world. GZIP is >> there to be used via HTTP. >> >> Kingsley >> > > As long as people know about it, it is okay. Hopefully the Linked Data > that DBpedia serves isn't limited in too many cases by this lack of > knowledge about having to use GZIP along with HTTP to resolve Linked > Data, even though it is widely available. > > Overall though, it is ironic that Linked Data is already running into > these capacity issues given its purpose is to get mass dumps of > everything that is known about single resources by performing a simple > HTTP URI resolution. Someone should implement a system that lets users > know what the limits are in RDF, and where they can go to > incrementally get more ;)... The issue won't just go away by telling > people to use SPARQL instead of Linked Data. > Peter,
I haven't had these restrictions applied on the basis "use SPARQL instead of Linked Data". They are applied because DBpedia is basically a very fuzzy project (its common definition and perception don't match *our* instance construction and upkeep reality), and I am now inclined to operate within *self imposed* narrower boundaries. DBpedia is comprised of the following (sadly unbeknown to most who assume its just about #1, but consume benefits of at least #1-3): 1. RDF Data Sets -- you can download these an load to an RDF store 2. SPARQL Endpoint -- the instance we host based on loaded Data Sets 3. Linked Data -- Resources bearing Structured Descriptions that are navigable via an HTTP user agent 4. Inference Rules -- OpenCyc, Yago, UMBEL, SUMO, DBpedia Ontology etc.. 5. Instance Maintenance -- 24/7 up keep and Bandwidth consumption etc.. So the bottom line is simply this, we are recalibrating what we offer (bearing in mind points 1-5). I also believe that gzip plus cache directives are more than ample re. DBpedia instance usage. If anyone needs to operate outside instance restrictions I've just outlined, they can simply contact us and we will consider making appropriate accommodations. Kingsley > Cheers, > > Peter > > -- Regards, Kingsley Idehen President & CEO OpenLink Software Web: http://www.openlinksw.com Weblog: http://www.openlinksw.com/blog/~kidehen Twitter/Identi.ca: kidehen ------------------------------------------------------------------------------ Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev _______________________________________________ Dbpedia-discussion mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion
