Re: [RIF][UCR] Suggestion for new Abstract UC

See below for comments.

public-rif-wg-request@w3.org wrote on 01/26/2006 09:43:40 AM:

> 
> Dear All,
> 
> John Hall wrote:
>
> ... stuff deleted
> 
> In my humble opinion, there are two opposite ways to achieve a language 
> for "Human-oriented Business Rules".
> 
> The first approach is to start with a natural language, say English, and
> 
> ... stuff deleted ...
> 
> The second approach is to start with a formal language,  and to
> derive from specifications (like eg: forall x y not (hand-over-to(x, y) 
> & intoxicated(x)) a rule in a so-called "controlled" natural language 
> (eg the rule in English John mentions.). This present approach presents 
> all promises and all difficulties of formal languages. Its advantage is 
> that its promises can be achieved today. Its drawback is that it is more 

> limited than natural language.
> 
> I believe that the W3C RIF WG should consider now the second approach 
> and keep the first approach for a future agenda -- maybe inb a few 
> decades or so. A controlled English  syntax for  RIF would perfectly 
> fit  in the  second work phase of the  W3C RIF WG.
> 

I strongly support this approach.  It is very important to separate out
the input/authoring aspects as well as as the output/display 
representation.

The first and most important step is to have a way of making precise 
statements.

This approach has been born out in practice by the MathML experience with 
many different
markups and markup tools.  Vendors have been happy to provide the 
mappings.

1. Provide a mechanism for defining objects and actions (or operators).
2. Provide a mechanism for explicitly tying such objects and actions to 
definitions.
3. Provide one or more ways of applying operators to objects.
4. Provide a default collection of definitions and as systematic a way as
possible of adding to or writing new collections.

Focus on communicating what is meant.  The author (authoring system) can 
decide
if the default definitions apply, and use them, or alternatively provide 
and refer
to enhanced definitions to clarify their intent.

The consumer (consuming system) can then determine if the constructed 
statement is
in its vocabulary or not.

Note also that there will be many different types of producers and 
consumers.
Imagine a natural language tool that generated RIF messages, or a RIF 
consumer
that verbalizes the RIF message it has received. 

Note also that not All consumers will Evaluate statements or compute 
outcomes.
e.g., An Editor that consumes RIF may simply be using it to build a more 
elaborate
message.

Stan Devitt
Agfa