Skip to main content.
home | support | download

Back to List Archive

Re: [SWISH-E:337] Re: Perl Processing of SWISH-E Results

From: Roy Tennant <rtennant(at)>
Date: Mon Jun 22 1998 - 15:48:43 GMT
Oops! That's what I get for messing around with it on the weekend with two
five-year olds running around the house. Those glitches are fixed now,
thanks. As for not doing the iterative search but scoring title hits
higher, I have two problems with that:

1) returning all of the hits at once can be a bit overwhelming, even if
the title hits sort at the top
2) with the directory grouping I'm doing, it wouldn't be intuitive to have
the title hits sort at the top and the rest sort below (the directory
would appear twice)

But all ideas are welcome. Thanks,
Roy Tennant

On Mon, 22 Jun 1998, Earl Fogel wrote:

> Hi,
> I took a look at this, and I am really impressed by the new output format.
> Sorting the results by server directory is an absolutely wonderful idea,
> although I noticed a couple of very small glitches in your implementation.
>     1) The search results page begins with:
> 	\nHere are the results of your search for <strong>search</strong>
>     2) The links to the directories don't work:
> 	<A HREF="http://Bibliographies/">
> I'm not so sure about the way you do successive searches (first title,
> then full text).  Wouldn't you achieve much the same results by adjusting
> the scoring algorithm to give more prominence to matches in the title?
> Earl Fogel
> Computing Services              phone: (306) 966-4861
> University of Saskatchewan      email:
> --
> On Thu, 18 Jun 1998, Roy Tennant wrote:
> >I've recently been changing how my site-wide searching works on the
> >Digital Library SunSITE, and since I've reached a bit of a plateau I
> >thought I'd share with you what I've done in case anyone finds it useful
> >(or entertaining <grin>). You can try it out at:
> >
> >
> >
> >The way it works is:
> >
> >* The first search defaults to finding words only in the titles of Web
> >documents; if that would fail, it automatically moves on to looking for
> >those words anywhere in the documents
> >* If the initial limit to titles would retrieve results, it stops and
> >displays them, along with a button which you can use to ask for more
> >results on the same search. If the user hits the button for more
> >results, the search words are searched across the full text of all Web
> >documents.
> >* The results are also sorted by server directory. The directory title is
> >translated to a more useful title than simply the directory name.
> >* The results of the second iteration are merged and displayed with no
> >duplicates, but items that were on the first screen are marked by a grey
> >bullet rather than the orange bullet which signifies items not previously
> >retrieved.
> >
> >I will let you see the (lousy) code if you promise not to laugh. Please
> >remember that I'm not really a Perl programmer, I just play one on the
> >Web. See:
> >
> >
> >
> >This is simply a symbolic link, so as I change the actual executable the
> >changes are automatically reflected in the program listing. If you have
> >any questions, let me know.
> >Roy Tennant
> >
Received on Mon Jun 22 08:56:05 1998