Skip to main content.
home | support | download

Back to List Archive

An idea for calculating ranking

From: koszalekopalek <koszalekopalek(at)>
Date: Thu Jun 09 2005 - 20:48:58 GMT
A couple of posts ago I was wondering how swish could get
a high/top ranking for when
searching for "documentation".  (That kind of enraged Bill
- sorry ;-)

Anyway, this is what I've been thinking about (at a fairly
abstract level):

1) The spider indexes document DDDD.

2) It locates an URL to UUUU and stores it (so that it can
    index it later on).

3) Instead of just remembering the URLs, let's also store the
    words WWWWW that are located inside <A HREF="UUUU"> </A>

4) When the url UUUU is indexed, add a meta name, say,
    "swishlinkedas" and assign value WWWWW to it.

5) Include the meta ("swishlinkedas") when searching

So, when indexing, I see a link to Inside the link to there is word "documentation".
I store this word in a hash indexed by the url.

When the spider eventually gets
it adds a meta "swishlinkedas" and assigns value "documentation"
to it.

One problem is that the spider knows all values for the
"swishlinkedas" meta only after it has spidered all documents.
But let's not worry about that now. Do you think the algorithm
makes sense?  I'm willing to do some work, err, some time during
holiday (unless the idea is totally brain-damaged)


Znajdz swoja milosc na wiosne... >>>
Received on Thu Jun 9 13:48:59 2005