A liberal arts student's research on PR algorithm _Shang pin China focuses on high quality website construction

The last advice is to cherish life, as a liberal arts student. Stay away from the formula!

All assumptions are based on user behavior analysis. The algorithm formulates these analyses. All algorithms have an assumption.

PR algorithm is mainly based on the quantity assumption and quality assumption

Quantity assumption: the more inbound links the page receives (links to the page from other pages are called inbound links), the more inbound links the page receives Website construction The more important. That is to say, a good page will definitely get many recommendations from other pages.

High quality web pages will transfer more weight through links. Quality assumption: the quality of inbound links pointing to the page is different. The more high-quality pages point to the page, the more important the page is. That is, a good web page will certainly be recognized by other good web pages.

At the beginning, PR algorithm will select a batch of web pages as seed web pages and give a higher PR to update the PR score of each page node through iterative recursive algorithm calculation, through the above two assumptions. Until the score is stable, it is the PR score of the current page.

As one of the factors of web page ranking. But PR is a global algorithm, and the result of PR calculation is the evaluation of the importance of web pages. It has nothing to do with the specific query, that is, the PR height that has nothing to do with relevance does not mean that the page is relevant. If the search engine only uses the PR algorithm to sort, no matter what query words you enter, the output results will be the same. Whoever has the highest PR will rank first.

Is PR important? Is PR not important? Look at the orchard! Excessive pursuit of PR results often outweigh the gains.

One high is not high, for PR. Everyone is really talented!

Suppose page A has two outgoing links that connect to page B and page C. If the PR value of page A is 1, the PR calculation is very simple. Both B and C pages will receive 0.5 value on average. This calculation method is based on the random walk model. The random walk model means that assuming that the web page has three outgoing links, the probability of users clicking each outgoing link is the same, so the PR value transmitted is the same

Otherwise, the PR value of all pages will be infinite. Therefore, PR algorithm introduces the concept of attenuation factor. Since pages are interconnected, PR cannot be transmitted circularly all the time. That is, the more times of transfer, the farther away from the seed page, the less PR value will be transmitted until the transmission value is 0 and the score is stable. The final PR score is calculated and added to the calculation of sorting results.

Some web pages are only in the chain but not out of the chain. Then the PR value of savings will become higher and higher and cannot be transmitted. This will violate the original intention of PR design and affect fairness. This structure is called a link trap.

That is, PR transmission is not limited to outbound transmission. Remote jump is a common way to solve link traps. PR can also be transmitted to any page with a certain probability

PR algorithm has been widely used in anti cheating for a long time, and is regarded as Google's landmark algorithm. That is to say, select a batch of cheating pages as seed pages (and also select trusted pages), give them a certain cheating score (or trust score), and transmit them like PR algorithm. Set a penalty threshold, and if it is reached, it will be a cheating page.

This anti cheating is based on the assumption that:

This page is probably also a cheating page. 1、 If a web page points its link to a cheating web page.

This does not mean that this page is cheating. 2. If a page is pointed to by a cheating page.

It depends on what problem this algorithm solves. Of course, this is just the most original anti cheating idea. Research on search engine algorithms should not only focus on formulas. Based on what assumptions, whether such assumptions are consistent with user behavior. Having solved the ins and outs of the algorithm, we can better know how search engines solve problems. This is learning SEO King way!

Prevent violations of rules from being punished. Solving rules is to better use rules.
This article was published by Beijing Website Construction Company Shangpin China //ihucc.com/

A liberal arts student's research on PR algorithm

Date：2013-06-18 Source: Shangpin China Type: website encyclopedia
Word Size： small medium big

+86 10-60259772