Thinking of Generalizablity

Although the page-rank algorithm works pretty well, it can not tackle with this condition, in which a malicious website put a lot of popular keywords on it to get a high priority when users search with multiple keywords.

To measure this kind of situation, I use generalizablity(概括度) to evaluate the results. A website tends to be a meaningless one with a high generalizablity

Definition:

Gene(web)=num(keywords)/con(w)   (a)

1 mum means number

2 con means content

If a website has more content and less keywords, it tends to have more details of a specific topic.

Extension:

The storage space of keywords is in directly proportional to num(keywords),so equation(a) can be:

Gene(web)=Space(keywords)/con(w)   (b)

The storage space of the web is in directly proportional to con(w) ,so equation(b) can be:

Gene(web)=Space(keywords)/Space(web)   (c)

Ok, we can easily work out the Gene easily.

Step1, make a statistics to the keywords in different fields

Step2, get the result of Gene with the Space of keywords in this website divide by the space of this web.