Review of RIF-FLD, version of April 14, 2008

Attached please find my comments on RIF-FLD,  version of April 14, 2008.

Comments are divided into general comments, specific comments, and  issues 
relating to clarity and writing style.

General comments: 
==============
Overall, this is a clear, good introduction to FLD. There are some parts, 
such as the discussion of signatures, that are not as clear as they could 
be. In several sections, there should be more motivation; discussion of 
how certain elements were introduced to enable and facilitate rules 
interchange would be helpful. There should also be more examples. This can 
certainly be published as a first working draft, but it would be good if 
the issues pointed out could be addressed in future versions.

Specific comments:
===============
Section 1, page 3, first paragraph: "It is a logic in which both syntax 
and semantics are described through a number of mechanisms that are 
commonly used for various logic languages, but are rarely brought all 
together."  It would be helpful to have a brief explanation of why you are 
all bringing all these mechanisms together --- because the basic framework 
must be broad enough to accommodate many different types of logic 
languages, because various sophisticated mechanisms are needed to 
facilitate translation into a common framework. 

p. 7,  item 8. in 2.1 :
There should be some discussion of rules vs. material implication. 

p. 7, end of section 2.1:

"A dialect might support all of these formulas or it might impose various 
restrictions. For instance, the formulas allowed in the conclusion and/or 
premises of implications might be restricted, certain types of 
quantification might be prohibited, classical or default negation (or 
both) might not be allowed, etc."

It would be very helpful here to have some examples.  Even  pointing to 
just BLD in this context (what does BLD have or not have?) would be 
useful.

p. 10, Section 2.4, definition of positional term.

"For instance, the above definition allows variables everywhere."

This is rather muddled. You want to say explain straight out the 
distinction between positional terms and  terms with named arguments. It's 
odd: of course, when you are defining terms in a regular logic, without 
terms with named arguments, it's sufficient to just give the general 
definition for positional terms without further explanation. However, once 
you do introduce terms with named arguments, you need to specifically 
explain what's going on with the standard terms.

p. 12, Section 2.5.

I found the discussion of external schemas confusing. It would be helpful 
to have an example of how you anticipate these being used. 

This was an issue that I found came up several other times in this 
document.  There are certain elements which appear to have been introduced 
based on how you expect the RIF to be used. But these elements aren't 
motivated, and aren't illustrated by example.

It will probably be quite a bit of work to do so. That is, the 
articulation of these examples is time consuming and will probably bring 
us several problems. But it really needs to be done to make this document 
clearer. 

p. 12, discussion of coherence for sets of external schema:  It would be 
helpful to discuss the intuition behind coherence. As it is written, I 
thought you might be characterizing sets of schema in which terms were 
unambiguous?

p. 14:  The discussion of the coherence of a set of signatures is not as 
clear as it could be. Again, a discussion of intuitions and motivation, as 
well as some examples, would help.

Example 1 on pp. 17-18 was clear and helpful.  Perhaps this example could 
be extended to address coherence.

I got what you were getting at with Example 2. I think the choice of 
Hamlet for this example is a mistake; see my detailed comments on this 
further down.

p. 19, section on EBNF grammar: This is a problematic section.  You say 
"The syntax of RIF-FLD relies on the signature mechanism and is not 
context-free, so EBNF does not capture this syntax precisely."

Well, more accurately, EBNF therefore doesn't capture the syntax. You say 
that the "EBNF grammar defines a strict superset of RIF-FLD." This is 
true. So what you really want to do is the following:

1. Define the superset-of-RIF-FLD class of languages that doesn't pay 
attention to signatures.
2. Present the EBNF for that.
3. Point out, as you already have, that therefore, the "EBNF grammar 
defines a strict superset of RIF-FLD." 

Otherwise you have a set of languages on the one hand, and an EBNF on the 
other hand, and neither one matches the other.

pp. 30--32, section on examples in XML serialization.

a. The example formalizations should first be presented in standard logic 
syntax, rather than just presenting them in XML. They would be much more 
readable and useful that way.

b. Also: I realize that using whimsical examples like your "translations" 
from Hamlet is common when trying to explain logic. I've seen this 
frequently done in logic textbooks. I think it's unfortunate, especially 
in a document of this sort, for two reasons:
First --- and I speak here as someone whose primary research interest is 
representing commonsense knowledge in formal logic --- it teaches people 
to construct terrible axiomatizations. You know that "Something is rotten 
in the state of Denmark" cannot really be translated as you have done in 
your example, by saying that there exists  something rotten that is in the 
state of Denmark. (The meaning is that there is internal corruption in the 
entity that comprises the *political* state. This is not trivial to 
formalize correctly) But someone reads this example, and the next thing 
you know, that person constructs an ontology of things, and one subclass 
is of rotten things, and another subclass is of things that don't love 
walls and send the frozen-ground-swell under them, and all our 
formalizations degenerate into silliness. 
Second --- because the example formalizations don't really capture the 
natural language, they're harder to understand.

Sometimes it really is better to be boring and humorless. Some sentences 
about pens of my aunt on the table would be fine. Even better would be 
simple examples from business, e.g., "Corporation X introduced a new 
plasma-screen TV in 2003. Sales topped $Y."


Issues related to clarity, writing style, typos, etc.
======================

p. 4, halfway down the page: "These terms are common in most languages." I 
think you mean "These terms are common *to* most languages."

p. 5,  2nd par in Signatures bullet: 
"For instance, the signature associated with a symbol, p, might allow p to 
appear ..."
This should read:
"For instance, the signature associated with a symbol p might allow p to 
appear ..."

If you include the commas around the first instance of "p", you are saying 
that p is the signature. As I understand it, however, p is the symbol.

p. 5, bottom of page:

"Roughly speaking, a set of formulas, G, logically entails another set of 
formula, g, ..."

a. You can leave out the commas around G or g. They don't change the 
meaning here, but they're unnecessary clutter. (And, I think, an actual 
error according to grammarians and style manuals.)

b. "Roughly speaking" should be omitted. There's nothing rough or 
approximate about what you're saying. It's clear that this is the natural 
language description rather than a formal presentation.

p. 6, top of page:

a.  "If I makes G true, then I also makes g true." 

"makes true" is an odd locution. Perhaps "If G is true under I, then g is 
also true under I."

b.  "Almost all known logics define entailment in this way."

Delete  the word "known".  What are unknown logics, precisely?

p. 19, top line:
"Up to now, we used mathematical English ..."
==>
"Up to now, we have used mathematical English"

What, in any case, do you mean by "mathematical English"?

p. 20, middle of page, item 5.
"each dialect is must define" 
==>
"each dialect must define"

p. 21, a few lines down:

"such as the well-founded negation"
 
omit "the"

p. 27, bottom line:
"The two most common theories of intended semantic structures are the so 
called well-founded models [GRS91] and stable models [GL88]."

"so called" should be omitted. I've seen this locution frequently, but it 
may not be the best expression here.  "So called" in English often has a 
connotation of illicit use, as if "well-founded" and "stable" are aliases 
for some other types of models trying to sneak in. (I have a feeling this 
locution may have crept in from an inaccurate translation from some other 
language.)