On Пнд, 2011-01-10 at 13:04 +0000, Toby Inkster wrote: > Don't we already have enough of those?
Maybe we do, but then it's an issue of "best practice" consensus and documentation. I think both publishers and consumers could benefit from a broadly-agreed-upon list of common link types with appropriate predicates. > <s> powder-s:describedby <o> . > # <o> provides some more information about <o>, often in a > # machine-readable format. Or could as well be a 15M PDF. An autonomous LD client cannot tell. An informal recommendation like "use wdrs:describedby to point to a terse and authoritative RDF description" and/or "wdrs:describedby is equivalent to a follow-your-nose description" would help. Then again, nothing in the POWDER spec says it shouldn't be a 15M PDF. So we kind of imbue a predicate with semantics it doesn't have according to the standard. A separate term with a clear specification would avoid this problem and everyone could rely on it. > <s> rdfs:seeAlso <o> . > # Like powder-s:describedby but perhaps more tangential. As Tim's email suggests, this interpretation is not universal. > <s> rdfs:isDefinedBy <o> . > # <o> provides the canonical definition of <s> . It's clear what a "definition" is for an RDFS vocabulary term, but not clear if I can apply this to instance data. Is "authoritative description" synonymous with "definition"? Does anyone use isDefinedBy for things other than classes and properties? -- Vasiliy Faronov