- From: Ivan Mikhailov <imikhailov@openlinksw.com>
- Date: Thu, 01 Jul 2010 20:49:33 +0700
- Cc: Linked Data community <public-lod@w3.org>, Semantic Web <semantic-web@w3.org>
Hello, Please, don't extend the existing model, for two reasons. >From implementor's POV, arbitrary literals are bad for any sort of indexing. >From AI specialist's POV, literals are simply not subjects. Can a number or a string _act_? Can you provide a living specimen of it? The feature is useless in modelling of the real world. so let it stay jailed in object position. If RDF could be extended, there are many better things to add, e.g., physical units. Finally, I don't want to get rounding errors on graphs like I can get them in CAD system. Should subject 3.14151956 and subject 3.1415195599999 connect their triples, can you guess? In any CAD I have a tolerance level to treat two points as "equal enough" and "their" geometric shapes connected; I'd prefer to never chose a tolerance level for a graph of heterogeneous or simply unknown nature. Best Regards, Ivan Mikhailov OpenLink Software http://virtuoso.openlinksw.com
Received on Thursday, 1 July 2010 13:51:27 UTC