From: | Robert Haas <robertmhaas(at)gmail(dot)com> |
---|---|
To: | Alexander Korotkov <aekorotkov(at)gmail(dot)com> |
Cc: | pgsql-hackers <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: Proposal: q-gram GIN and GiST indexes |
Date: | 2011-04-05 13:05:07 |
Message-ID: | BANLkTi=AUAJxcDFGcScTijne2SkHb3p77A@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
On Tue, Apr 5, 2011 at 8:41 AM, Alexander Korotkov <aekorotkov(at)gmail(dot)com> wrote:
> For example, here is distribution of q-grams count in 120 Mb of dblp paper
> titles (pretty large dataset).
> q count
> 2 7218
> 3 115107
> 4 589428
> 5 1648453
> 6 3336685
> Number of 5-grams if about 15x larger than number of 3-grams. But most part
> of index space will be occupied by links to the rows(about 120 millions of
> links), while size of q-grams itself will be almost ignorable in comparison
> with it.
I am probably being stupid here, but doesn't the number of links to
rows grow proportionately to the number of n-grams?
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company
From | Date | Subject | |
---|---|---|---|
Next Message | Luca Ferrari | 2011-04-05 13:30:10 | Re: 9.0.3 SIGFAULT on FreeBSD with dtrace |
Previous Message | Alexander Korotkov | 2011-04-05 12:41:34 | Re: Proposal: q-gram GIN and GiST indexes |