Re: Selectivity

Originally posted by Gorang
> On Wed, 30 Apr 2003 10:33:07 +0000, andrewst
> wrote:
>
> >A query has "highly selectivity"
> Would that rather be one of the following:
> A query is highly selective
> .. or ..
> A query has high selectivity
> ?
>

It would.

Originally posted by Gorang
> There is no contradiction.
>
> Yes there is! :)
>
> Selectivity is defined as a number.
> This number's low values corespond to 'high selectivity' and high
> values to 'low selectivity'.
> This implies that the value of a term does not correspond to the
> meaning of a term. Which, IMHO, is contradiction.
>
> For instance zero selectivity is highest possible selectivity, or is
> that another exception?
>

And the no. 1 record has the highest chart position...

But perhaps the number should be called something else like "hit ratio". Selectivity is high when the "hit ratio" is low and vice versa.

Originally posted by Gorang
> P.S. This whole issue is actually of minor importance and might be
> caused by the fact that english is not my primary language. All I
> wanted to know is does this fall into blunder of a definition category
> or should I try to improve my english (or logic)?
> ( GoranG79 AT hotmail.com )

I'd say this was more to do with English than logic. The English word "selective" implies rejection: the more "selective" something (e.g. a query) is, the more things (e.g. rows) it rejects, and the less it accepts.
The more records a query SELECTs, the less SELECTIVE it is!!!

The number used to indicate selectivity is "the wrong way round" (if you like) because it seems more intuitive to say: "this query returned 10 out of 1000 records, so selectivity = 1%" rather than "this query failed to return 990 out of 1000 records, so selectivity = 99%

--
Posted via http://dbforums.com