- From: Barclay, Daniel <daniel@fgm.com>
- Date: Wed, 8 Apr 2009 11:46:01 -0400
- To: "Dan Brickley" <danbri@danbri.org>
- Cc: <public-swd-wg@w3.org>
- Message-ID: <49DCC6B9.80107@fgm.com>
Dan Brickley wrote: > On 7/4/09 21:22, Barclay, Daniel wrote: >> The term skos:broader should be named with something less ambiguous >> than the >> single word "broader." ... >> The name should contain something that indicates the direction of the >> relationship (the way "subclassOf" uses the word "of," or something like >> "hasPart" uses the word "has). > > Actually I think this was a regrettable mistake in RDFS. We really > should have used the names "superProperty" and "superClass" for those > relationships. > > The 1st RDF spec said > http://www.w3.org/TR/1999/REC-rdf-syntax-19990222/ > > Note: The direction of the arrow is important. The arc always starts at > the subject and points to the object of the statement. The simple > diagram above may also be read "http://www.w3.org/Home/Lassila has > creator Ora Lassila", or in general "<subject> HAS <predicate> <object>". > > > This pattern is pretty common in RDF. We write "age", not "hasAge" > usually. Note that "broader" doesn't actually follow that pattern: That pattern involves nouns, but "broader" is a comparative adjective. That difference causes it to not fit the pattern: trying to read it as "<subject> HAS <broader> <object>" doesn't succeed at making sense of it. (Arguing for following that pattern (not the only solution) would argue for renaming "broader" to "broaderTerm": the phrase "broader term" is a noun, so "<subject> HAS <broader term> <object>" works perfectly. That also matches your preferences for "superProperty" and "superClass.") The pattern that the adjective "broader" does suggest is the one for "greater"/"greater than", "older"/"older than", etc.: Consider "<x> <greater> <y>." Would you think it meant x > y ("<x> IS <greater than> <y>") or x < y (presumably via taking the original as "<x> HAS <greater [value]> <y>")? Presumably your answer is "x > y". Imagine how confusing it would be if the predicate <greater> were defined such that "<x> <greater> <y>" mean that x was _less_ _than_ y. That would cause no end of confusion. That's how the current definition of skos:broader appears to me. Also, note how we don't usually say just "x greater y"; we say at least "x greater than y." That would argue for the solution of renaming "broader" to "broaderThan" to be clear (and reversing the sense, of course). > But that said, I agree that "broader" takes a bit more thought than > other properties. I think this is because it is a property whose range > and domain are the same. "parent", "sibling" etc are similar in that > regard. No, it's not because the domain and range are the same--it's because there's a big difference between nouns and comparative adjectives, in particular, in how they imply relationships. Imagine filling out a form about a person or a taxon X. If you saw a form box (or a data field) labeled simply "parent," you'd probably think it was asking you for the object of "<X> has parent <O>," probably not asking "is person or taxon X a parent?" and surely not asking for the subject of "<S> has parent <X>." However, a form box labeled "broader" would likely make one think "broader than what?" a lot more frequently than suggesting to enter a broader term. On the other hand, a label of "broader term" very likely would accurately make one thing it was asking for term that is broader than X. Daniel -- (Plain text sometimes corrupted to HTML "courtesy" of Microsoft Exchange.) [F]
Received on Wednesday, 8 April 2009 15:46:53 UTC