Re: ternary search trees question
On 9 Aug 2005, at 13:08, Tone =8Akoda wrote:
>
>
> posted by: Tone =8Akoda <tskoda at email.si>
>
>
>
>
> Maybe some expert on this topic will be able to answer this question:
>
> What if we insert key "aba" and then "abandon". Since data for key is sto=
red in
> eqkid, "aba" can't have any data because eqkid will be required to store
> "abandon" key.
>
> Am I missing something?
>
> I'm writing ternary search tree database because creation (not access) of=
> database with EDS is too slow for something I'm doing.
>
> http://algorithm.myrice.com/resources/technical_artile/ternary_search_tre=
e/terst
> ree.htm
The part i hope someone can explain to me is this: how are the branches of=
such trees ordered, on what criteria, and how is such pre-set-up criteria=
going to be optimal for a search in any but for the pre-set-up question?
For instance, you can set up a tree for words who's every other letter is t=
he
same, or every 2nd letter is the same, or have common endings (minus the=
suffixes), or have the same 3rd letter. But, for instance, how is the ranki=
ng for=20
3rd letters arranged for "cow" vs "cot" vs "pow" determined, by what criter=
ia?=20
Why would "pow" be anywhere near "cow"? I understand pattern matching,
and i understand hashing, but to pre-set the database for fastest searching=
of=20
a pattern, you have to know in advance what that pattern is! This leads to=
many databases optomised for many patterns.
Kat
|
Not Categorized, Please Help
|
|
