Search Strategy Product Manager Must-Read Series—Lecture 5 Page

Before officially introducing the Page. Rank algorithm let’s start with a case in real life. Everyday when we write papers we often cite other people’s. Papers and classic papers in a certain industry will be  by a large number of papers. If the paper happens to be by another classic paper. It will highlight the importance and authority of the paper even more. In fact, the same is true for the importance and authority of web pages. Therefore, we set the following two assumptions. Quantity assumption: When a webpage is  to by other webpages more, and the number of links is larger, the webpage is more important.

Basic Algorithm of Page Rank

As shown in the figure above, the website is cited by many websites and links are which means that the website is very important. Quality assumption: When linked to by a high-quality webpage, it means that the linked webpage is also of high Iran Phone Number List quality and authoritative. As shown in the figure above, the website is cited by the high-quality website , forming a link, indicating that the website is also authoritative. The overall idea of ​​the PageRank algorithm is based on the above assumptions. . Basic Algorithm of Page Rank Based on the above two assumptions, we introduce the Page Rank algorithm. First of all, we imagine the Internet as a graph network. Each node (Node) of the network is an independent webpage.

Specific cases

If there is a hyperlink relationship between two webpages, there is a directional edge between them Edge the number of nodes that each node links out is  the out-degree of the node. The Page Rank value hereinafter  to as PR value of each node indicates the authority of the node. Our core is to build a user’s walking model in the graph network, and update and iterate the PR value based on the walking model. The above is the basic BY Lists definition of the Page Rank algorithm. First, the PR value of node is  by the PR values ​​of other nodes linked to this node. Suppose the linked nodes are The more other nodes are  the greater the PR value of the node, so the algorithm iteratively.