How do SE determine if one site is related to another as far as the cotent is concerned? I see people looking for related links but some are really picky and other not so... If I take sport for instance, it's obvious that two sites about golf are related. Bot what about sport sites in general? Lets say a golf site and formula 1 site? Is a basketball site link to water sport site link any better at all than a link from a mobile phones site?
I use common sense first of all, and second I use the online tools for related researches such as th one you can find at http://www.digitalpoint.com/tools/suggestion/. There's one on Google if you have a Google Adwords account. They should give you a good idea of what people look for other than the main keyword. But common sense is the best criteria always...
I believe that Google is now able to pick up words such as "golf" and "Formula one" and say.... right, both are sports, therefore these are connected.
mightyb; IMHO I am not so sure the se's can make that connection, the really rely on the appreance of words apprearing in natural language. Think of it this way - if the se had to take the English SAT tests they would not do very well.
Yes the search engines can see the difference, not because they see a link between formula 1 (which could be a shampoo) and golf. But both sites will likely have other related terms like sports, athlete, in this case "drivers" ect. these additional terms will make the sites related even if they are not with the main keywords of the site.
G only looks the words . If there are same words on both web site, they are related according to google.
I've been reading up on google and the recent changes, they are attempting to vary the way the se views linked sites and are providing new tools, but how effective that is I'm not sure. For example an adult dating site and another dating site not adult will like adac.. says have other similar words and phraseology. Consider the converse why would 2 sites not be linked there are bound to be words present on both sites that match so the technology is more complex having to look at phrases and types of words. As an example consider doing a search would you want to search the world leaders or world leaders either way I believe the search engines view this the same by ignoring the word the. My understanding is that it works in a similar way when linking similar sites. Links is another way they look at who is linking to you. If all links are from the same source or category in directories tehn the chances are the sites are similar or same subject. Its all based around stats and number anyway but I know they keep changing, "improving"
Google uses LSI - it's fascinating stuff. Take the time one day to do some research about it. There is a trick to highlight what Google thinks is related and it's very simple actually. For example: Google things Dogs and Cats are related. But they don't think that horses are related to dogs and cats.