- From: Giovanni Tummarello <giovanni.tummarello@deri.org>
- Date: Wed, 19 Dec 2012 19:27:10 -0600
- To: Vishal Sinha <vishal.sinha777@yahoo.com>
- Cc: "public-lod@w3.org" <public-lod@w3.org>
Generally speaking its yet another big gaping unsolved problem. Our stab at it one is to do big data Hadoop based summarization and then use the summary to understand how to query it. E.g. this unpublished link below exposes the Sindice.com data graph summary containing a summary of 20-30 billion triples in sindice. http://demo.sindice.net/dataset/ take a look at the bbc example, you can see classes, for each class you can see which property you can use etc. In this way it becomes possible to write queries that make sense and not in the blind. An application that uses this graph (this sort of graph) is then the assisted sparql query editor (you point it to the summary graph and have these recommendations) http://sindicetech.com/sindice-suite/sparqled/ (Contains simple description of the data graph summary) Gio On Wed, Dec 19, 2012 at 12:28 PM, Vishal Sinha <vishal.sinha777@yahoo.com> wrote: > There are many public SPARQL endpoints available. > For example the cultural linked data: > http://cultura.linkeddata.es/sparql > > How can I know what type of information is available in this dataset. > Based on what assumption I can query it? > Do I need to know any structure beforehand? > > Viashal >
Received on Thursday, 20 December 2012 01:27:58 UTC