Search module gone mad? Possible Bug?

[Russ]

OK I'm building a site and testing the search module which is providing eratic results.
If I search for "N1 Exhaust Muffler" (sans quotes)

I get:
Search Results For "N1 Exhaust Muffler"

    * PS Revolution Exhaust Muffler - PS Revolution Exhaust Muffler (100%)
    * N1 Evolution Exhaust Muffler - N1 Evolution Exhaust Muffler (84%)
    * GT Spec Exhaust Muffler - GT Spec Exhaust Muffler (84%)
    * N1 ExTi Exhaust Muffler - N1 ExTi Exhaust Muffler (84%)
    * N1 Universal Exhaust Muffler - N1 Universal Exhaust Muffler (76%)
    * N1 Exhaust Muffler - N1 Exhaust Muffler (76%)
    * ATS (Active Tail Silencer) - ATS (Active Tail Silencer) (61%)
    * N1 Dual Exhaust Muffler - N1 Dual Exhaust Muffler (46%)"
    * Super Silencer - Super Silencer (23%)

Note the order and %'s even though a page with "N1 Exhaust Muffler" exists and is listed?
This page has a title of "N1 Exhaust Muffler" and a menu text of "N1 Muffler" and page is here:
http://www.dp.shoesforindustry.net/prod ... hausts/n1/

Search is here:
http://www.dp.shoesforindustry.net/home/search/

If I search for "N1" which 'seach' should find pages for (see above) I get no results??

What does the search module 'search' on - any idea why the odd results. I have cleared the cache of the web site and my browser and also re-indexed the search?

Any help appreciated...
Russ

[Piratos]

If I search for "N1" which 'seach' should find pages for (see above) I get no results??

code:

Code: Select all

	// ignore 1 and 2 letter words
	if (strlen($word) <= 2)
	  {
	    continue;
	  }

you  cannot get results with words <3 chars only you change the code.

------------------------

You see the big diff from Time Taken to the time over all ?

Search Results For "N1 Exhaust Muffler"
Time Taken: 0.109551

[Russ]

Thnaks Piratos, that explains the no-find with 'N1', but it does not explain the other bizzare results?

Anyone have any other ideas?

Russ

[Russ]

MMmmmm, I've now added piSearch to the page as well for comparison. Neither seem to give the result I was expecting?
http://www.dp.shoesforindustry.net/home/search/
Russ

[Piratos]

N1 Exhaust Muffler

Is this the Title {title} within the text or it is a h1 tag ?

Pisearch works in the content not in the  title (but i can change it).

[Piratos]

Download module pisearch 1.52 , i have included the title in the search.

[Piratos]

Interesting  the built in finds pages and show it if the settings of the page is show_in_menu = false;

[Russ]

Thanks Piratos, I think the internal search is meant to do it this way - for 'no-mennu' pages perhaps like 'Privacy Policy' where you might not want ot show it on the menu but want to search for it?

Also thanks  for the latest version of pisearch, I've put it up but it still seems wrong to me. I guess I'm thinking 'google' like in my search. ..
So I would award max % or weight for a page title then it's h1, h2. h3 and then content etc.? Does this seem right to anyone else. If this were the case, in my example at the start of this post, then the 'N1 Exhaust Muffler' page - it's in the title, would be at or near the top as one might expect if searching for it!! Indeed I'd be unahppy if it wern't and it would make a bit of a nonsense of searching?

Can anything be done to the internal search or pisearch to fix this? It strikes me that this is how it should be? What does anyone else think?

Update:
Actually it is worse than I thought with Internal Search. I changed the page we are searching (only temp - it's back now) to have the same menu and title "N1 Exhaust Muffler". Even with this it still only ranked second on the list depsite having a title and menu the same as the search phrase? This cannot be right can it?

Russ

[Piratos]

The order of the searchresults with Piserach is the hierarchy and so you get never a result with "N1 Exhaust Muffler"
at the first position if there are another pages with the same content at a better place.


But my interest is this in your Testsite:

Built in shows the Time Taken: 0.04956

But the page is generated complete :

Generated in 2.765934 seconds by CMS Made Simple using 29 SQL queries

You must know that Pisearch  parsed all contents with smarty, if a left and right delimiter ( {} ) was found.
But than you can find all content that comes from modules and plugins that is visible in the pages.

At this view the Pisearch - Time is very good

Generated in 3.704906 seconds by CMS Made Simple using 40 SQL queries

and - if i disable with Pisearch the Smarty parsing the time in summary is better than the built in search.