- From: Thomas Passin <tpassin@tompassin.net>
- Date: Mon, 18 May 2020 17:36:08 -0400
- To: Cory Casanave <cory-c@modeldriven.com>, Patrick J Hayes <phayes@ihmc.us>
- Cc: thomas lörtsch <tl@rat.io>, Semantic Web <semantic-web@w3.org>
On 5/18/2020 2:01 PM, Cory Casanave wrote: > ... Agreed, but included to complete the requested "semantics of graphs". The discussion about first/next, etc. seem to be more a data structure approach where as "ordered closed set" more semantic. It seems to me that the simplest thing that could work would be to give each thing in a collection (one that is intended to be used for a list) a total order (or maybe a partial order, but that starts to get more complicated) property. Software that knows about this property can sort the items, and ordinary RDF software will be able to tolerate this property without any new syntax or semantics - they just won't be able use it for sorting or concluding anything about the list items based on their order property. RDF inferencing would remain the same as it is - sorting using the property would take place outside any standard RDF inferencing. The property wouldn't affect open- or closed- world assumptions since it would only be used for ordering that particular collection outside of any RDF inferencing. Yes, most RDF software wouldn't be aware of the total order property, but they aren't really good at handling order anyway, or this discussion wouldn't be going on. As an aside, I don't see list ordering as being a good candidate for adding to RDF anyway, because that order isn't unique: there is an large or indefinite number of ways to order a given list, so any one specific ordering is not really a property of either the list items or the list itself. It is only a transient snapshot of someone's (sorting) preference at a point in time. TomP
Received on Monday, 18 May 2020 21:36:24 UTC