in many sites we see the recommendation function, also use some sites in seeking love or Shanghai script pendant, however recommend out of the article but there are two problems: the correlation is not big, cannot arouse the readers’ interest.


through different search terms of different users in different articles, and these search words in other users is similar.

article I have seen many blogs out of the code, this method should be better, but the problem is that many bloggers on the choice of the label is not professional label selection is often arbitrary, even if the two articles contain "love Shanghai" label, may article is about the change of love Shanghai search rules, and another with sea in annual revenue, two articles have no obvious correlation.

1. recommended Pendant

artificial intelligence to understand.

crawler statistics and cluster databaseThe computer has

through historical records to determine some related articles, many users click to enter an article, enter another article at the same time, the strong correlation.

is a little difficult to understand, we can only illustrate by examples: A, B, C three users. A, B, C three articles. A search for "W" click into the a article, B search for "m" click into the B article, C search for "P" into C, Google W, M P, found that these three phrases have very similar meanings, so a, B and C determine the three articles are related articles, so below the three article returns a list of related articles (JavaScript calls).

Look forward to more powerful Google

in this area, although our expectations of love Shanghai, but I actually believe that Google’s technology. The principle is very simple:

According to the label recommended

so, is there a way to improve the recommendation agreement? The answer is yes.

if you have to rely on Google search keywords, it is too silly, through Google statistical code, from browsing history to carry out in-depth judgment. When A entered the a article, found beneath a random list of recommendations, if found to have their needs or articles of interest to B, will click in, these actions were recorded on Google big data, when the a enters B by innumerable people after practice, even for B in which the a pages are more likely to click on, many times the time of entry are not willing to click the B article, and Google have the ability to do this analysis, finally, when B C first entered the A, the B link will appear in the very reasonable position. In fact, if millions of users from a into C, and finally to this article, why not directly to the C article on the a page? This is somewhat the meaning of big data.

Leave a Comment

Your email address will not be published. Required fields are marked *