In order to assign the relevance of a word in a web page, a statistical analysis has been done on the words used in several languages.
This was performed with a modified web cruncher. The result were 7 files with the word frequency for English, Arabic, German, Spanish, Japanese, Portuguese and Russian languages.
For instance, these are some frequencies of English, Japanese and Spanish words took from their respective files,
These numbers are the global frequencies of a word. Afterwards, a web cruncher will calculate the local frequencies of words in a particular web page.
The relevance of a word in a particular web page is the ratio of its local frequency to its global frequency.
local frequency word relevance = ----------------- global frequency
If a word appear more frequently within a web page than it appears in all web pages then it will have ratio larger than 1 in that web page.
With button audit you could know the relevance of words in a web page according to this simple algorithm and is the one used by Trokam to rank pages.