[thelist] PHP Search Engine

Pringle, Ron RPringle at aurora-il.org
Thu Mar 16 15:06:38 CST 2006


Excellent point. Using the Zoom engine there are ways to weight the
search by adding meta tags to your pages, but of course that is a pretty
involved effort on even a medium sized site like ours, and I have only
used it to a limited degree.

Another problem we run into is the precedence of PDFs over actual HTML
content. Again, it can be solved by giving the HTML content by weight,
but again, lots of effort.

Paul, I'd be interested in the results of your research, feel free to


p.s. and thanks for the tinyurl for that link! ;)

> Hi all,
> Just a note to BEWARE of a LOT search engine software. I've 
> been researching a replacement piece of software for our site 
> off and on for the past few months and have found that much 
> of the software on offer (free and paid) just isn't very 
> effective in returning good results.
> Sure, it may spider your site. Yes it may build a lovely wee 
> index database, but the search results are often atrocious.
> Case in point: - (No, I'm *not* picking on Ron here, just 
> illustrating a rant er... Point)
> http://tinyurl.com/o56lx
> This is a search on the site Ron mentioned, for a piece of 
> text that appears as an H2 in a plain html page.
> Results: 0
> An acid test for all search software is to grab some text 
> from the vendors site and see if their own site search finds 
> it. You will be amazed....
> Additionally, as we've found with our awful site search 
> software, it _just_ _doesn't_ _scale_. We have a site with 
> around 1Gb of content - mainly static HTML and PDF. The site 
> is reindexed nightly (no incremental indexing - that broke 
> within weeks of the install) the index db is almost twice the 
> size of all the actual site content and the indexer refuses 
> to drop old pages. The search is slow, poor quality and I 
> can't wait to kick it's sorry butt to the curb.
> Some software may work well with 10's or 100's of pages, but 
> when it gets to the 1000's it's a whole different ball-game.
> If anyone wants the results of my research so far, let me 
> know. I may save you some time
> Paul

More information about the thelist mailing list