- From: Michael Hausenblas <michael.hausenblas@deri.org>
- Date: Wed, 18 May 2011 08:57:45 +0100
- To: Enrico Franconi <franconi@inf.unibz.it>
- Cc: W3C RDB2RDF <public-rdb2rdf-wg@w3.org>
Enrico, I very much appreciate your input. Now, I think I start to understand what we shouldn't do, but I don't understand what we should do, in your opinion. Can you please come up with a concrete proposal how to handle this? Cheers, Michael -- Dr. Michael Hausenblas, Research Fellow LiDRC - Linked Data Research Centre DERI - Digital Enterprise Research Institute NUIG - National University of Ireland, Galway Ireland, Europe Tel. +353 91 495730 http://linkeddata.deri.ie/ http://sw-app.org/about.html On 18 May 2011, at 08:45, Enrico Franconi wrote: > On 18 May 2011, at 06:52, Juan Sequeda wrote: > >> Seems like an unofficial consensus is to completely ignore NULL >> values. > > This is also not information preserving wrt the normative behaviour > in SQL. > This shows up when there is an update: you can update a null value, > since it is compatible with the schema; on the other hand, you can > not update a missing value if there is no attribute in the schema. > The NULL values also affect aggregated queries, so they can not be > ignored. > The SQL way is to introduce a special NULL constant *and* requiring > that the queries introduce an inequality test for all join variables > (which should never be equal to NULL). This introduces no additional > cost in terms of query answering complexity (indeed SQL NULL values > are efficient); it is rather expensive in terms of reasoning over > queries, e.g., query containment (see [1]), since it introduces > inequality in the queries. > cheers > --e. > > [1] Containment of Conjunctive Queries over Databases with Null > Values, by C. Farré and W. Nutt and E. Teniente and T. Urpí; <http://dx.doi.org/10.1007/11965893_27 > >.
Received on Wednesday, 18 May 2011 08:04:05 UTC