Skip to main content.
home | support | download

Back to List Archive

Re: Ranking Question

From: Bill Moseley <moseley(at)not-real.hank.org>
Date: Tue Dec 17 2002 - 21:08:59 GMT
On Tue, 17 Dec 2002, Nikolaus Rath wrote:

> I have some questions regarding the ranking of search results:
> 
> - Does the context of the match influence its ranking? (e.g.,
>  "<h1>match</h1>" ranks higher than "<p>match" when searching for
>  "match"). And if so, which tags are used?

Yes.  It's a hole in the documentation, it seems.

  a tag of h* where "*" is any digit get's the IN_HEADER flag set.
  tags <em> <b> <strong> <i> gets IN_EMPHASIZED flag set.
  <title> gets IN_TITLE set.
  <head> gets IN_HEAD, <body> gets IN_BODY.

  <meta> and xml tags and "fake" tags in the body of html get 
  IN_META flag. 

Then in config.h there's settings:

  #define RANK_TITLE      4
  #define RANK_HEADER     3
  #define RANK_META       3
  #define RANK_COMMENTS   1
  #define RANK_EMPHASIZED 0

> - How does the ranking work in general?

Search the list archives.  Here's one:

  http://swish-e.org/archive/4232.html

See also rank.c.  You should probably look at it in CVS:

 http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/swishe/swish-e/src/rank.c?rev=1.9&content-type=text/vnd.viewcvs-markup

> - Does swish-e search the given pattern also in all the meta tags? If
>   so, does a meta match count more than a "normal" match?

Yes.  


-- 
Bill Moseley moseley@hank.org
Received on Tue Dec 17 21:09:18 2002