[pcdb] some ideas for data modelling
jedd at progsoc.org
Fri Mar 30 22:12:17 EDT 2007
On Saturday 31 March 2007 6:35 am, paul at heliosville.com wrote:
> But -- oh crap -- what about the fact that you can have multiple parentage,
> and hence multiple paths?
Hierarchy just isn't going to map well (or rather, allow reality to
I had a thought this morning regarding guilds -- do they need to
be explicitly stated, or can they be deduced from more general
IOW, rather than have a walnut/currant/hackberry relationship
explicitly stated somewhere, could the system come to the conclusion
that those things work well together from simpler information.
Consider a scale of cohabitation of -10 (bad) through 0 (neutral)
to 10 (good).
F.e. the following facts are recorded within a db:
walnut coexist well - say 8 - with hackberry (bi-directional)
hackberry coexist well - say 8 - with currant (")
currant coexist well - say 8 - with walnut (")
walnut coexists badly - say -8 - with apple (")
walnut coexists neutrally - 0 - with black walnut (")
You say you have a walnut, and ask the system what guilds
work well with this tree. It goes through and looks at the
coexistence (or guild potential) of each entity, and reports
1. walnut, hackberry, and currant should work well together
2. walnut and black walnut do not compete but don't offer
any benefits to each other
3. you should avoid planting a walnut near an apple
Obviously this needs a level of abstraction for varieties within
species, and some weighting aggregation based on the number of
people who've asserted 'hackberry coexist well with currant' and
the weightings of how well each person's stated they coexist, etc,
but as an overall concept .. ?
I like the idea of not recording guilds as a separate or discrete
entity unto themselves, and can't think of any situations where
the above approach would fall down.
More information about the pcdb