Skip to main content.
home | support | download

Back to List Archive

Stemming problem, I think

From: Robert Stoeber <Robert(at)not-real.usabusiness.net>
Date: Thu Dec 29 2005 - 02:39:59 GMT
Hi,

I'm running Swish-e 2.4.3 under Red Hat Linux and it's been working
great.  We have a site with about 40,000 press releases in MySQL and
most everyone is very happy with the search results.

However, one company that's not happy is called MicroE Systems.
When I search for MicroE the results are a bunch of documents
containing micro and micros, but MicroE is far down the list.  Why
doesn't the exact match rank highest?

Even putting quotes around the name as "MicroE Systems" gives me
other companies called Micro Systems apparently ranking higher than
the string I searched for.

MicroE is found eventually and highlighted in the results so I know
the documents are being indexed.

One thing I tried is raising the rank value for the company name by
putting it in a Meta field like this:

MetaNamesRank 10 company

The documents themselves contain this:

<html>
<head>
<title>$headline</title>
<meta name="company" content="$companyname">

That didn't seem to make the documents I want rank any higher.

The configuration file contains a line like this which I assume
controls the stemming:

FuzzyIndexingMode Stemming_en2

I'm not sure what's going on or how/if I can fix this problem.  Like
I said before, we've been happy with the search results until now
and I'm not sure what to do about this.
Received on Wed Dec 28 18:40:20 2005