Forum Moderators: martinibuster
Here I'll attempt to dimistify the concept in hope of making it used more frequently in its proper meaning and not abused like it now commonly is, where you find newbies refering to everything as a filter and any effect is there "because of an xyz filter".
The main thing to know about a filter is this: A filter is a Yes or No function, it is a True or False, it is a black or white, it is pass or do not pass, it is belongs to or does not belong to (in math sets languge).
A good example of a filter is the one used to automatically put spam into your Bulk, Junk, Spam (whatever they like to call it) folder at your email box. This is indeed a filter. As a message passes through it, it can say Yes or No. It can either say this is an acceptable message or a Spam message (and thus automatically put it in the junk folder for you, if it has been configured to do so).
So, the lesson is: filters say either Yes or No, they do nothing other than this. How do filters work is another issue beyond this post (email filters usually use a bayesian algorithm to learn from members what is and what is not spam, this is quite simple method to do machine learning and is not resource-demanding like other AI techniques are).
As for search engines, do they use filters? Well, yes. How about a filter for scraper sites that aim at spamming search engines? That's a good thing to creat a filter for. As email spam filters are not perfect, so are spam site filters in search engines not perfect. In theory they should get better with time as they learn more from examples. That's one reason why it is a good idea that you submit to Google any spammy site you discover, because it helps their system learn about patterns common in such sites and enhance it's spam-sites "filter".
That was a good example of a filter that search engines would like to use. It's a filter because it says either yest or no, and that's it.
As for other things that search engines determine for a site such as its ranking or the like, some have refered to such things as a filter for sites that are new, a filter for sites that get too many links in a short period of time a filter for ... just about anything. Such issue sure are dealt with in search engines, but it is more of a penalty mechanism that uses variables to add up (or take from) the overall formula that calculates the metrics of a web page (or web site). Such things cannot be described as filters, because their answer is not just a harsh yes or no answer, but a variable value that enters into play with other variables that calculate the overall metrics of a site or web page.
