I've used some of the RIFRAF classification to put together a preliminary
taxonomy of features that are of interest to the RIF. While this pictorial
taxomony is much easier to grasp than the RIFRAF description, I realized
that drawing it is very hard because of the inevitable clutter. I had to
omit a number of combinations of various features, which make sense both
semantically and syntactically. (This doesn't mean that RIF should endorse
all combos, but having a comprehensive diagram could be useful.)
I used 2 kinds of arrows. One denotes extensions that add expressive power
and one that doesn't. I avoided calling them "semantic" vs. "syntactic"
extensions, because these things are not well-defined.
In one case I used a third kind of arrow (extension of a sublanguage)
because I had to cut off a subset of Datalog that uses only binary and
unary predicates in order to reduce clutter (SWRL is sitting directly on
top of this).
I also tried to place various languages at their appropriate places in this
taxonomy.
I was fuzzy about the "fuzzy" extensions, because there are several, and
also vague about what is meant by NAF (there are 2 most popular ones and
several others).
This is all very prelim and I am a bad diagram drawer. Hopefully someone
here is better at this. I am attaching a png, which can be used in any
browser; and SVG file, which can be edited and displayed in the browser
with an appropriate plugin (despite the claimed Firefox 1.5 support for
SVG, it can't display that one); and the source file for the DIA tool,
which I used to draw all that.
--michael
