Falko Braeutigam wrote:
On Thu, 05 Apr 2001, Roy Ladner wrote:Isn't data mining relevant to both extremes? Wouldn't data mining in an Ozone environment be useful for the highly structured data? Is there anything in Ozone that would hinder data mining?
>
>
>
> Falko Braeutigam wrote:
>
> > On Thu, 05 Apr 2001, Azhar Hussin wrote:
> >
> > > have anyone ever use ozone for data-mining purposes ?
> > > I am thinking to do that
> >
> > In general, ozone is great to run applications with highly structured object
> > graphs. It is not that suited to run applications with large numbers of uniform
> > data. SQL is better for that.
> >
>
> Can you explain what you mean by 'large numbers of uniform data' and why ozone is
> not suited to that sort of application.millions of data records without any structure (relations). consider a
sales database of Amazon or something. Just item num, prize, date, etc. No
structure. You just need to store the data. Efficient searching needs very good
strategies here. RDBMS are great for that.On the other hand consider a database of molecules. Each atom only has up
to say 10 connections to other atoms but there could be really lots of them in
there. Highly structured. Such applications need a way to model this
structure in a natural way. RDBMS are unable to do this. Searching atom with
weight X (what an RDBMS could do) makes no sense here, because the structure is
importantOf course, these are the two extremes. You need to check for each project
what exactly does this mean 'large number' and 'highly structured'.
Dr. Roy Ladner
Naval Research Laboratory