lancis
Elite Member
- Jul 31, 2010
- 1,682
- 2,443
@tar729 cant say I understood this post, but the other one was convincing. 
I asked for some math on your other post, and would love to see some math here is well.
Here is why:
1. Google/Yandex/Whatever rely on algorithms, algorithms rely on math. If something cant be formulated mathematically, it means only one thing - it doesnt exist in the world of search engines.
2. On the grand level of things everything boils down to the problem of estimating the probability of page given a keyword, e.q. p(page|keyword). The higher the probability the higher the page is in the search results. There might be variations, such as p(page|keyword,country), p(page|keyword,user), but these are only subproblems of the main problem above.
According to Bayes: p(page|keyword) = p(keyword|page) * p(page) / constant
p(page) is the well known pagerank algorithm
p(keyword|page) deals with semantics and relevant links, and it would be great if you could shed a light on how exactly p(keyword|page) is estimated by Yandex. If you can, give us math instead translations an we will derive the structure.
I asked for some math on your other post, and would love to see some math here is well.
Here is why:
1. Google/Yandex/Whatever rely on algorithms, algorithms rely on math. If something cant be formulated mathematically, it means only one thing - it doesnt exist in the world of search engines.
2. On the grand level of things everything boils down to the problem of estimating the probability of page given a keyword, e.q. p(page|keyword). The higher the probability the higher the page is in the search results. There might be variations, such as p(page|keyword,country), p(page|keyword,user), but these are only subproblems of the main problem above.
According to Bayes: p(page|keyword) = p(keyword|page) * p(page) / constant
p(page) is the well known pagerank algorithm
p(keyword|page) deals with semantics and relevant links, and it would be great if you could shed a light on how exactly p(keyword|page) is estimated by Yandex. If you can, give us math instead translations an we will derive the structure.