Re: racer documentation

So if I understand correctly, RACER uses typing on datatype properties
in order to keep the different datatypes separate. This means that you
can combine ontologies without harmful interaction between different
datatypes if the signatures are disjoint, i.e., if there is no
datatype property that is used by the two ontologies in an
"incompatible" way. So the situation is analogous to the issue with
transitive roles and cardinality restrictions.

Another issue is that Racer indeed does not allow multiplication, so
Hilbert 10 is not an issue. From the same manual:

--------------------------------------------------

RacerPro supports reasoning over natural numbers (N), integers (Z),
reals (R), complex numbers (C), and strings (S). For different sets,
different kinds of predicates are supported:

N | linear inequations with order constraints and integer coefficients
Z | interval constraints
R | linear inequations with order constraints and rational coefficients
S | equality and inequality

--------------------------------------------------

greetings,
 		Carsten


On Fri, 7 Dec 2007, Jeremy Carroll wrote:
>
> Looking at the racer user manual v 1.9.2
>
> I found
> [[
> 3.6 Concrete Domain Attributes
>
> Attributes are considered as "typed" since they can either have fillers of 
> type cardinal, integer, real, complex, or string. The same attribute cannot 
> be used in the same T-box such that both types are applicable, e.g., (min 
> has-age 18) and (>= has-age 18) are not allowed. If the type of an attribute 
> is not explicitly declared, its type is implicitly derived from its use in a 
> T-box/A-box. An attribute and its type can be declared with the signature 
> form (see above) or by using the KRSS-like form 
> define-concrete-domain-attribute. If an attribute is declared to be of type 
> complex is can be used in linear (in-)equations.
> However, if an attribute is declare to be of type real or integer it is an 
> error to use this attribute in terms for nonlinear polynoms. In a similar 
> way, currently, an attribute of type integer may not be used in a term for a 
> linear polynoms, either. If the coefficients are integers, then cardinal 
> (natural number, including 0) for the type of attributes may be used in a 
> linear polynom. Furthermore, attributes of type string may not be used on 
> polynoms, and non-strings may not be used in constraints for string
> ]]
>
>
> Jeremy
>
>
>

--
*      Carsten Lutz, Institut f"ur Theoretische Informatik, TU Dresden       *
*     Office phone:++49 351 46339171   mailto:lutz@tcs.inf.tu-dresden.de     *

Received on Friday, 7 December 2007 10:51:20 UTC