Re: Weighted Searching - Mailing list pgsql-sql

From Josh Berkus
Subject Re: Weighted Searching
Date
Msg-id 39BE8933.85875E2@agliodbs.com
Whole thread Raw
In response to Weighted Searching  ("Mitch Vincent" <mitch@venux.net>)
List pgsql-sql
Mr. Vincent,

> I emailed the list a while back about doing some weighted searching, asking
> if anyone had implemented any kind of weighted search in PostgreSQL.. I'm
> still wondering the same thing and if anyone has, I would greatly appreciate
> a private email, I'd like to discuss it in detail.. I have several ideas but
> most of them are pretty dirty and slow..

You really need to do this in PLSQL, Perl or C because there isn't any
good way to implement weighting in pure SQL -- weighting is a
procedureal thing.

I'm currently in the process of designling an HR app that will use quite
elaborate weighting scheme.  Candidates are compared against job
openings and the matches are weighted according to the degree of
similarity (e.g. Job A requires 5 years of experience and the candidate
has 4 or -20% for a weight of -1 but he is in the right location for a
weight of +2 etc.) with the summary of weights to be fudged by the user
according to what s/he considers most important for the job (location,
skills, etc.

I'm not done but I expect this function to be 4-5 pages of PL-SQL. 
Fuzzy logic eats processing power.  I'll be open-sourcing the program in
December, but you probably can't wait that long.

I'd be interested to hear from anyone who's written a web search
engine.  It seems to me that the algorithm for, say, google should be
fairly similar to what I'm doing for HR.
                -Josh Berkus

-- 
______AGLIO DATABASE SOLUTIONS___________________________                                       Josh Berkus  Complete
informationtechnology      josh@agliodbs.com   and data management solutions       (415) 436-9166  for law firms, small
businesses      fax  436-0137   and non-profit organizations.       pager 338-4078                               San
Francisco


pgsql-sql by date:

Previous
From: Jie Liang
Date:
Subject: Odd stuff
Next
From: Frank Bax
Date:
Subject: Re: making two columns out of one