- From: Bob Morris <morris.bob@gmail.com>
- Date: Fri, 24 May 2013 17:26:39 -0400
- To: public-openannotation <public-openannotation@w3.org>
OK, I understand that the graphical conventions [1] in the Feb 2013 spec are purely informative, and about the doc, not the model. But I have come to decide that in publications it would be helpful to stay as close to the graphics of the spec doc as feasible. In attempting to do so, I have---so far-- one minor whine and one whimper. Whine: I see no powerful reason to distinguish object predicates' arrows from datatype predicates' arrows by whether they are straight or curved. That distinction should be guided by the layout constraints in the diagram. For both cases, it is easy to find oneself forced into graphical ugliness by sticking to the convention. Among the consequences are wasteful extra white space needed when straight arrows are "required" and silly trivial curves when curved ones are "required". By "required" I only mean, "attempting to follow the documentation conventions." Besides all that, the two types are already distinguished by the shape of their targets---ellipses for objects, lozenges for literals. So it would be nice if the documentation were simply silent on the matter of curves or straight arrows. It wouldn't even require changing the pictures. Whimper: It's a stretch to say "Class instantiation (rdf:type) is depicted as a straight black line with white arrow head, following UML." Most (all?) UML systems produce diagrams in which the equivalent of rdf:type is denoted with ":" in the label of a box depicting the object and its properties. There are no arrows for this particular UML association between instances and classes. Other associations between any UML objects are denoted with solid lines, and, if those associations are directional, with open arrowheads. Open arrows are generally (always?) reserved for the "generalizes" association, useful in specifying a class hierarchy. I like the triangular headed arrow convention for rdf:type, because it's an easy to find graphical signpost that let's you ignore the predicate name when it doesn't matter. But crediting it to UML makes me more, not less, confused, especially in the face of other UML reserved words, like "association" that are also used differently in UML. I would remove "following UML" from [1]. Bob Morris [1] http://www.openannotation.org/spec/core/index.html#Examples -- Robert A. Morris Emeritus Professor of Computer Science UMASS-Boston 100 Morrissey Blvd Boston, MA 02125-3390 IT Staff Filtered Push Project Harvard University Herbaria Harvard University email: morris.bob@gmail.com web: http://efg.cs.umb.edu/ web: http://wiki.filteredpush.org http://www.cs.umb.edu/~ram === The content of this communication is made entirely on my own behalf and in no way should be deemed to express official positions of The University of Massachusetts at Boston or Harvard University.
Received on Friday, 24 May 2013 21:27:14 UTC