Re: Merging Rulesets

Gerd Wagner wrote:

> Coming back to your scenario, I think it's realistic to
> have the following kinds of rules:
> 
> a) pimozide is contraindicated with macrolides according to a 
>    1996 FDA bulletin 
> 
> b) pimozide is safe in conjunction with macrolides for men
>    over 60 according to a 1999 FDA bulletin

[An excellent example.]

> Then b would logically contradict a, and we would need
> a nonmonotonic conflict resolution procedure such as
> giving higher priority to more specific and/or more 
> recent pieces of knowledge.

Or we might decide that medical decision making was too important to base 
on generic conflict resolution procedures and instead require someone to 
explicitly resolve the interaction between the rules. In that case the 
ability to detect the contradiction would be useful, indeed perhaps a 
requirement.

By the way, this use case as currently phrased doesn't actually seem to 
strictly require merging (in the sense of one ruleset using the results of 
another). The case comprises:
  a. data on Bob's prescriptions
  b. the ontological data that "erythromycin is a macrolide antibiotic"
  c. a rule-based "encoding of the 1996 FDA bulletin..."

I seem to be missing something but I see only one ruleset plus ontology 
data (presumably encoded in OWL).

I can see that there might be multiple sources of drug-interaction rules 
for (c) but I would have thought they could be kept separate, or at least 
not require non-trivial merging. I don't want a drugs company's interaction 
rule to override or affect an FDA rule.

I assume the idea is that there is information in category (b) that is 
needed for the drug-interaction rules to usable but which can't be encoded 
in terms of data and ontologies using RDF and OWL. Does someone have an 
example? [Not an example of information that can't be encoded in OWL, I can 
come up with lots of those in general, but an example which fleshes out 
this particular use case enough to be a convincing example of why ruleset 
merging is needed here.]

Dave

Received on Tuesday, 30 August 2005 10:00:14 UTC