[Data-modeling] Reworking the Structure type

Paul Houle paul at ontology2.com
Tue Feb 24 15:39:31 UTC 2009


Iain Sproat wrote:
> Over the past day or two spencermountain and I (sprocketonline) have 
> been drafting up some proposed changes to the Structure type in the 
> Architect commons.
>
    Funny,  I recently noticed that the "Nuclear Power Station" type was 
pretty weak (it seemed mostly to be a rant about the fact that a power 
station produces waste that has to go somewhere) and added my own type

http://www.freebase.com/view/user/paul_houle/default_domain/nuclear_reactor

    which covers research reactors,  power plants,  etc.   I thought 
about making "Structure" an included type of a Nuclear Reactor,  but it 
just seemed to me that I'd never fill in most of the fields in Structure.

    I also created types for "Nuclear Reactor Coolant" and "Nuclear 
Reactor Moderator".  I would have been happy if there was a "Substance" 
type that would include both chemical compounds (water) and chemical 
elements (sodium) but didn't see one.

    I created a "Nuclear Reactor Type" field that is a tag-like type 
that can be used in several ways:  for instance,  some types describe 
the purpose of a reactor "Research Reactor",   and others describe the 
major design ("Light Water Reactor",  "High Temperature Gas Cooled 
Reactor") of the reactor.  In a lot of cases,  the choice of Coolant and 
Moderator could be specified in the type (for an LWR or CANDU reactor 
for instance) in other cases it wouldn't be: "Breeder Reactor" is a 
statement that the conversion ratio > 1,  not a statement about the 
design or materials.

    It particularly bugs me that I'm not representing the hiearchy of 
reactor types...  I'd really like the system to know that

Light Water Reactor -> Pressurized Water Reactor -> GE System 40

    so I can just say "GE System 40" and the fact that that is an LWR is 
inferred.  Of course,  real life hierarchies tend to be imperfect.  for 
instance,  the government is now pushing the idea of "Gen 3",  "Gen 3+" 
and "Gen 4" reactors and you can't really say that an LWR is 3 or 3+ 
(old ones are 3,  new ones will be 3+) and I don't know if I'd want to 
say that an experimental breeder that was built in 1957 was "Gen 4".

    Any thoughts?



More information about the Data-modeling mailing list