Skip to main content.
home | support | download

Back to List Archive

RE: Questions re Swish-E and Search Results

From: Bill Moseley <moseley(at)not-real.hank.org>
Date: Tue Jul 02 2002 - 22:51:52 GMT
At 03:16 PM 07/02/02 -0700, Hall.Liz wrote:
>When a user types in a term such as "gene" (for example)in the search box, a
>list of urls are returned and there might be a description with "gene"
>highlighted. However, when you click on the link to the file or web page, it
>seems as though you are always taken to the top of the page and not to the
place
>in the file where the word "gene" is located. Is there any way to do this?

In general, you are searching for documents about something, and don't
really care where your search words are in that doc.

An exception to that would be if the documents are large and what you
really want is for search results to take you to the section of the
document that matches your query.  An example of swish doing that is at:

       http://perl.apache.org/release/

Type is, say "zombie fork" and the first hit is:

   mod_perl: Performance Tuning: Avoiding Zombie Processes

and it says:  Page: 333.0k, Section: 6.7k which mean the page is 333k bytes
long, but the section for that specific hit is 6.7k bytes.  Clicking on the
link takes you to that specific section.

>I understand you could have anchors in the page, but this is
>not really practical for the type of web page I wish to index (since it is a
>list of journals/databases).

Well, it doesn't have to be anchors, it can be any HTML markup that you can
identify.  But it has to be something.

I suppose you could pass to the page the query word and use Javascript to
jump to the right spot.  But I'm not sure how you might deal with a complex
query.  The java search function would have to work in a way similar to the
way swish-e searches.


-- 
Bill Moseley
mailto:moseley@hank.org
Received on Tue Jul 2 22:55:19 2002