- From: Joshua Allen <joshuaa@microsoft.com>
- Date: Tue, 20 Dec 2005 13:01:25 -0800
- To: "Henry Story" <henry.story@bblfish.net>
- Cc: "Frank Manola" <fmanola@acm.org>, <semantic-web@w3.org>
> All I see is a re-statement of an opinion on your litmus test page. > Not a backing of it. That is correct. Frank asked for clarification of my opinion about rdf:type, and I believe that clearly lays out my opinion. As for backing it up, I wasn't intending to get in a debate about it. My predictions are pretty clear, and history will tell if I was right. To clarify a bit, I am basically saying that "type information gets in the way of broad semantic interoperability". Or "the first wave of semantic web will be built with the type of information which does not require strong typing". Another way to look at it -- I think software should be written to look for the specific predicates that it needs, without regards to the rdf:type. The "type" is defined by the existence of a collection of predicates. By introducing rdf:type, all you do is introduce the possibility that the rdf:type and the collection of predicates might mismatch, and you foil merge scenarios. Then you end up with different engines all deciding which route to take (honor rdf:type, or honor the predicates which you can plainly read?). My opinion is that the meaning of the predicate (or combination of predicates) should be clear enough that it does not require additional information to be processed. Using rdf:type as a hint is fine, but it should always be possible to process the information in the absence of rdf:type. I realize that there are many scenarios where this will not work. Those scenarios will not be mainstream v1 semantic web scenarios.
Received on Tuesday, 20 December 2005 21:02:26 UTC