Re: Polygon data modelling question

>paul_geoffrey_brown_at_yahoo.com (Paul G. Brown) wrote in message >news:<57da7b56.0201172215.32623052_at_posting.google.com>...
>
> I've read the VA file papers. Two observations.
>
>
> Hope this adds some fuel to the fire.
>
> KR
>
> Pb

Well, VA-file is not a shoe fit-for-all-size. An obvious problem is that if the query radius is bigger than the width of its slices, then it is no use to do the filter step. About the k-NN and range query, each one has its usage in the different applications. For instance, the former is always useful in feature comparation algorithms, while the latter is always useful in the geographic information systems ... This problem is too complex to be easily stated here in a short essay since hundreds of papers in database, cs theory can't clarify it ...

So for this problem discussing in the application context, maybe no conclusion is the best conclusion. I should say - the applications are so different, that best way to know which index is better is testing and use real data and real query do testing.

Thanks,

Qingqing Received on Sat Jan 19 2002 - 02:53:40 CET