From: Jeff Heflin (heflin@cs.umd.edu)
Date: 01/09/01
Frank, The last two weeks have been really hectic for me, so it isn't until recently that I had a chance to give the Walkthrough and Reference a detailed look. Sorry they're so late, but here are my comments: Reference document: ---------------------- 1) Imports: I think we need to explain to the reader why we have imports, and what it gives us that namespaces don't. That is, the namespace rec does not imply anything about definitions of terms, and even if we say that it should for DAML, there are a number of of valid ontologies that will provide new definitions without introducing new names (they add to the definitions of terms provided in other documents). Theorectically, we could augment the import mechanism to do what namespaces give us (since, if you import definitions, you obviously mean to import the names they apply to as well), and get rid of namespaces, but since we live in XML world, I guess we're stuck with them. 2) Property Elements: Don't UniqueProperty and UnambiguousProperty specify global cardinality restrictions? That is, no matter what class the property is applied to, the cardinality constraints must hold, unlike the various "cardinality" properties used in a Restriction. I think we should be clear about the differences between these two types of cardinality restrictions. 3) Appendix - List of language elements: It would be nice if this was sorted alphabetically and hyperlinked to the appropriate descriptions of the elements in the document. That way, someone can easily look up what a particular element means. Walkthrough document: ------------------------- 1) Housekeeping: We might want to say a little bit about why we need imports, as I mention above. 2) Defining properties: We might want to say what it means to have multiple domain restrictions 3) Defining properties: (concerning comment about better alternative for domain and range restrictions). After having thought about this a while, I don't think we should necessarily say the use of "Restriction" and "toClass" is better than "domain" and "range." I think there are pros and cons to both and we should simply present these to the reader so he/she can decide on a case by case basis. Pros for domain and range include ability to be grokked by normal RDF agents, and the fact that sometimes maybe you really do want to restrict the property globally. 4) Defining property restrictions: "Such an assertion about the class Animals has exactly the SAME STATUS as the assertions made within..." I think we should clarify that if such assertions are made from another document, then they only have the same status if the other document is equally trusted! 5) Notations for properties: See my comment for the reference about UniqueProperty and UnambiguousProperty specifying global cardinality restrictions. We should probably point this out here as well. 6) Notations for properties: "... not yet that Ancestor is the transitive version of hasParent ... We would need to introduce additional elements to enforce these connections." This means that DAML doesn't yet have the expressivity for this, right? If so, can we make that a little clearer? If any of these are controversial, maybe we can discuss during the telecon today. Jeff
This archive was generated by hypermail 2.1.4 : 04/02/02 EST