- From: Pete Rivett <pete.rivett@adaptive.com>
- Date: Thu, 18 Nov 2010 11:21:45 -0800
- To: <public-rdb2rdf-comments@w3.org>
My area of concern is the treatment of null values. By not generating any RDF for null values you are changing the original semantics. Though, admittedly, the semantics of null in relational databases are not universally clear (and much discussed/disputed), I would expect some attempt to map it. At the minimum a discussion of the issues. In the example Sue has an address of NULL. That could be interpreted as the positive assertion that she has no address i.e. she is homeless or 'with no fixed abode'. The RDF in effect states that "we know nothing about Sue's address" which is quite different. In fact the RDF is even different from the main alternative relational interpretation of NULL which is "Sue's address is unknown" i.e. it's a "known unknown" which is different from a complete absence of knowledge about it. Subtle but different. Regards Pete -- Pete Rivett CTO, Adaptive Inc
Received on Thursday, 18 November 2010 21:47:43 UTC