Welcome! Log In Create A New Profile

Advanced

Proper conversion for weights

Posted by Chukillz 
Proper conversion for weights
August 11, 2011 11:22PM
Hey guys i need some help....what would be an easy way to calculate your weights so that they work best?

is there some sort of math formula to calculate this?

what are some suggestions that could help me figure out what my weights should be set at please?

thank you in advance

Chukillz DaKlown
Tec
Re: Proper conversion for weights
August 13, 2011 03:24PM
The weighting of each keyword is calculated in the script
…/admin/spiderfuncs.php
See row 827 – 729 for the algorithm.
Even if the function calc_weight() is (falsely) commented like:
//function to calculate the weight of pages
In reality this function calculates the weight of each keyword.
The algorithm contains some multiplier called:
- $title_weight
- $domain_weight
- $path_weight
- $meta_weight
These variables are presenting the values you may define in admin backend by the settings called:
- Relative weight of a word in the title of a webpage
- Relative weight of a word in the domain name
- Relative weight of a word in the path name
- Relative weight of a word in meta_keywords

In order to understand the algorithm, which calculates weighting of each keyword for original Sphider, it might be helpful to write it in another way:
    $weight =  ( ( $words_in_page
                        + $word_in_title * $title_weight
                        + $word_in_domain * $domain_weight
                        + $word_in_path * $path_weight
                        + $meta_keyword * $meta_weight
                  ) * 10
                  / (0.8 + 0.2*$path_depth)
               );
Now you will clearly notice that modifying the settings in Admin backend alters the multiplier in the corresponding row. But the main structure and variables of the algorithm remain unchanged.

Now, if you intend to vary the Admin settings, you should keep in mind some additional info.
The sum of all variables defined in Admin backend should not exceed a value of 100
Take notice that a keyword found 100 times in full text ($words_in_page), which even has no multiplier that you could modify, will always be weighted stronger than a word found in "title". Consequently the influence of modifying the variables in Admin settings is limited.
Another argument you should not miss is the additional Admin setting called:
- Keyword weight depending on the number of times it appears in a page is capped at this value
This setting could be used to limit the count of a keyword. Capped at this value depending on the number of times it appears in a page, so that the weighting will not exceed for a page.
This is a powerful setting to influence the weighting of keywords in full text.
In other words, you may influence weighting by modifying the Admin settings. But whether the result listing afterwards will become better, clearer, more objective, or more subjective?

By the way, if you like a more flexible weighting, Sphider-plus is offering additionally
- Promoted keywords
- Promoted domains
- 9 methods how to sort the results
http://www.sphider-plus.eu/index.php?f=14#14_7

Tec
Re: Proper conversion for weights
August 13, 2011 07:29PM
Hey tec i have sphider plus but the only thing wrong with it is that i didnt pay for it...i got it from the forums here under the GPL...which means i have no support for it...which means all the codes i have are for regular sphider sad smiley how ever i do plan on paying for shpider plus so that i can get the mods and fixes from the forums on sphider plus...

Now with my next question is this... instead of putting in the code from the script to show me where to look for it...could you possibly help me out by looking at how i have my weight settings and showing me how i would figure this math problem out that you showed in your post please...

Weights
20 Relative weight of a word in the title of a webpage
60 Relative weight of a word in the domain name
10 Relative weight of a word in the path name
3 Relative weight of a word in meta_keywords

These are my settings...

Thanks Tec

Your Friend
Chukillz
search algorithm
March 06, 2014 04:44AM
which search algorithm used in sphider search engine,
i need name of the algorithm used,
thank you.
Tec
Re: Proper conversion for weights
March 06, 2014 12:46PM
The name of the algorithm is 'Ando-Saabas-search-algorithm'. In other words, Ando prepared his own algorithm. Don't try to google for the algorithm name, it is his individual solution for his crawler. I just created this name without asking him. Hopefully he doesn't mind.
In case you like to understand the algorithm, try to follow my explanations at http://www.sphider.eu/forum/read.php?2,8636,8643#msg-8643
Yes, you are right. This is part of the index procedure! But searching for a keyword will always be the result of proper indexing.

Tec
Re: Proper conversion for weights
March 07, 2014 04:36AM
hello tec,
thank you for your help.
Re: Proper conversion for weights
March 20, 2014 04:40AM
hello,
which algorithm used in sphider search engine for indexing
may i know algorithm name.
thank you
Tec
Re: Proper conversion for weights
March 20, 2014 01:04PM
The name of the index algorithm is 'Andos-best-knowledge-how-to-index-algorithm'. Any further nonsense question?

Tec
Re: Proper conversion for weights
November 19, 2017 12:49PM
hello,
which algorithm used in sphider search engine for weighting

thank you
Tec
Re: Proper conversion for weights
November 19, 2017 07:47PM
http://www.sphider.eu/forum/read.php?2,8636,8643#msg-8643

Tec
Sorry, only registered users may post in this forum.

Click here to login