Re: Demo: Modelling Cost of Travel Paths Between Towns
Date: 18 Nov 2004 11:34:43 -0800
Message-ID: <4b45d3ad.0411181134.4b1933c6_at_posting.google.com>
> > Mary likes John
> >
> > Assuming user already knows John, Mary, and likes, the above is
> > the correct way for a user to express/view the relation ...
>
> What does it mean to say "[a] user already knows ... likes."
I may have made a poor assumption. I assumed the human mind works more like TM/XDb2 then RM because a mind demonstrates extreme flexibility to represent things. In TM/XDb2, things (ie John, Mary, likes) must already be represented before creating the relation "Mary likes John" because the relation refers to those things (ie ->Mary ->likes ->John) rather than storing the originals in the relation. Storing a thing itself in the relation (ie Mary likes John) typically leads to data-dependent refs which can lead to inflexibilities especially in AI-type apps, for example not being able to handle two towns with the same name as in Celko's original RM schema, because no one specified two towns couldn't have the same name before design-time. If you don't think a human mind has to already represent John, Mary, likes before making a relation involving them, I am not about to argue this point. Take the blue pill.
> If each representation can only appear once in the computer,
> we can't ever mention anything twice,
After representing a thing once, use a data-independent ref to the original thereafter. Below I mention John twice without representing him twice:
John // One and only representation of John ->Mary ->likes ->John // A relation which does not metion John // but refers to John.
Below is how it can appear in XDb2 Script:
CREATE *; // this thing represents john
CREATE it.cls = person;
CREATE it.name = john;
CREATE *; // this thing represents mary
CREATE it.cls = person;
CREATE it.name = mary;
CREATE *; // this thing represents like
CREATE it.cls = verb;
CREATE it.name = like;
CREATE mary.like = john; // Under the hood ->mary ->like ->john Received on Thu Nov 18 2004 - 20:34:43 CET