[Corpora-List] Corpora and SQL

[Lars Nygaard]

Lars Nygaard wrote:

> 
> This works, but slowly. It's better to use two temporary table
> 
>  drop table temp_table_a;
>  drop table temp_table_b;
> 
>  create temporary table temp_table_a (x int primary key) type=HEAP;
>  create temporary table temp_table_b (y int primary key) type=HEAP;
> 
>  insert into temp_table_a(x) select ID+3 from corpustable where word = 'a';
> 
>  insert into temp_table_b(y) select ID from corpustable where word =
> 'como';
> 
>  select temp_table_a.x from temp_table_a,temp_table_b where
> temp_table_a.x=temp_table_b.y;

Also, I guess I should mention, to get a satisfactory performance, you 
probably need to create some variants on this scheme based on the 
absolute and relative frequencies of the tokens you are searching for.

cheers,
lars nygaard