Re: Special query type for Spatial DB

"Mikito Harakiri" <mikharakiri_at_iahu.com> wrote in message news:<BXUNb.8$KI3.136_at_news.oracle.com>...
> Your question has many interpretation. What is the universe of discourse? Is
> it a metric space, or graph? How do you define "closeness" metric for a set
> of points? Do you want the exact answer, or approximate one?

It can be metric-space (with some additional work added to support it).
Closeness = shortest distance
i.e. For every point in a given neighborhood of points, its distance to every other point is the shortest.

Approximate is fine.

> Assuming that you can get away with approximate answer, could I suggest
> density map? Density is a function that associates a number with every point
> in your metric space, or graph. Density map is an incremental evaluation
> structure on your universe of discourse. Every time you modify your
> universe -- add a pulsar point into 3D space, or add a fire station node
> into network of existing firestation -- you have to update your density map.
> Quering density map is easy: just fimd me the maximum.

Density map sounds very interesting.
However, I see a problem in adding more to the WHERE-clause if my query is slightly modified so that it becomes: Find me 5 fire-stations in California that are closest to each other and everyone of them must have, at least, 2 free fire-engines.

Is there an easy way to augment a density-map with other properties, like the free fire-engines example so that it can support more complex queries?

Good discussions, guys! Received on Tue Jan 27 2004 - 19:32:49 CET