Posted by Brian A. Anderson on 01/19/05 01:58

Quoting: "Richard Lynch" <ceo@l-i-e.com>




> Unless you KNOW your site will get loads of traffic from the get-go, I'd
> focus more on the SIMPLE solutions that you can maintain as a beginner.

Quoting: "Richard Lynch"

> Unless you KNOW your site will get loads of traffic from the get-go, I'd
> focus more on the SIMPLE solutions that you can maintain as a beginner.

well, I have implemented a simple solution, and it is working, but I am
thinking in terms of in the coming year or two of developing more
sophisticated strategies.


> Also depending on the size and scale and scope of your data, I'm not sure
> that two separate lists is a Good Idea.

Why not? I have arount 1000 products and many searches bring up more than a
few pages worth of data. I would think that it might be good to present the
links to test files along the side.

Here is a link to the actual site:
http://www.apsistl.com/

> Amazon may need it for, their zillion items they sell, but do you, really?
>
> I've spent *DAYS* writing special search engines for data-sets that only
> had 100 items in them, and would never grow significantly larger than
> that. I told my client that was pretty wasteful of their money, but
> that's what they wanted. [shrug] Pays the bills. But that doesn't mean
> you want to waste your money/time that way.

Right now the people visiting my website are complaining that when they
search for "sports cars" for instance, that they get many sports items
sorted first before they even get to see the "sports cars". I realize that
this is a separate issue, but it illustrates that even though my search is
working, it could be improved. I look forward to figuring out how to do it
better. Regardless, my data sets may not always be limited to 1000 products
(not to mention the text pages).

> One Axiom: Keep as much of the scoring/sorting in your SQL as possible --
> That's what SQL engines are best at.

I agree. That is a wise statement, and I have tried to do this with regular
expressions but this doesn't deal with relevance at all, or is there
something that I am missing? I suppose I could do multiple searches on my
data to look for "sports cars" first, "cars" second, and "sports" last. Then
I could deal with the separate issue of the text files.


-Brian

[Back to original message]