Chapter 3825 hits algorithm

He waved to the staff next to him, and soon 5 whiteboards were carried to the reporting platform.

They prepared a lot of whiteboards, originally intended for Qin Luo.

Unexpectedly, it was used on Jon Kleinberg first.

"Thank you." Jon Kleinberg thanked, and then looked at everyone: "Before I formally demonstrate, I want to tell you what the HITS algorithm is."

"This is a brand-new calculation mode. When the user enters a keyword, the algorithm calculates two values ​​for the returned matching page, one is the hub value, and the other is the authority value. The hub value refers to all the keywords on the page. The sum of the authority values ​​of the pages pointed to by the outgoing links. The authority value refers to the sum of the hubs in the pages where all the incoming links are located..."

"Compared with the traditional pagerangk algorithm, the HITS algorithm can better describe the organizational characteristics of the Internet, and the convergence speed is faster, reducing the search time..."

"..."

Eloquently, Jon Kleinberg briefly explained the advantages of the hist algorithm.

There is a lot of content, but it can be summed up in two sentences.

The HITS algorithm reacts faster and takes less time.

After a brief introduction to the HITS algorithm, Jon Kleinberg began to make a formal report.

"I think of it this way, submit the query q to the retrieval system based on keyword query, take the first n web pages (such as n=200) from the set of returned result pages, as the root set, record it as S, then S satisfies :

"Low number of web pages in 1.S"

"The webpages in 2.S are the webpages related to the query q"

"The web pages in 3.S contain more authoritative web pages"

"..."

"By adding the webpages cited by S and the webpages referencing S to S, expand S into a larger set T. Take the Hub webpage in T as the vertex set V1, and take the authoritative webpage as the vertex set V2..."

"The hyperlinks from the web pages in V1 to the web pages in V2 are edge sets E, forming a bipartite directed graph...."

"Where I operation: a(u)=Σh(v); O operation: h(v)=Σa(u). Each iteration normalizes a(u) and h(v)..."

"..."

At first, everyone listened with gusto.

But as the argument deepened, everyone's eyes began to become more and more confused.

They actually understood a little bit.

But the crowd didn't show the slightest contempt, instead the expressions on their faces became more and more serious.

Who is Jon Kleinberg?
A world-renowned computer scientist, he is now a professor at world-class Cornell College.

Will he talk nonsense?
Of course not.

Since there is no problem with Jon Kleinberg's argument, there is only one answer. The reason why they can't understand is not because Jon Kleinberg is talking nonsense, but because the knowledge he expresses is too profound , so that they could not understand for a while.

It is precisely because they can't understand that they feel that Jon Kleinberg's argument is correct.

Jon Kleinberg's reporting continues, and there's no limit to stopping.

"What I just said is the specific algorithm of the HITS algorithm."

"Next, I'm going to talk about the pseudocode in the algorithm..."