HillTop Algorithm in SEO Google algorithm parsing Series

In the previous article, we introduced the potential semantic index (LSI). Today we will introduce the peak of hyperchain analysis:HillTopAlgorithmAs one of the core ranking algorithms of Google, there is a lot of literature on the Internet. This article focuses on the analysis of the original algorithm, without considering too many complex factors, so that you can better understand the nature of the algorithm.

HillTop algorithms combine PageRank, HITs, and Correlation Algorithms. Krishna Bharat at the Compaq System Research Center and George A at the University of Toronto. mihaila filed and applied for a patent in 2001, and then authorized Google. In December 2003, the Google algorithm was updated and became one of Google's core ranking algorithms.

HillTop is a query correlation link analysis algorithm, which overcomes the disadvantages of PageRank's query independence. Simply put, the HillTop algorithm is an algorithm that sorts search results based on popular query keywords. ForPopular keywordsThis is because the HillTop algorithm runs less efficiently. The algorithm mainly consists of two processes:

I, Expert pageSearching and scoring. After finding popular keywords based on user query logs, the search engine starts to search for expert pages for these popular keywords, which becomes two necessary factors on the expert page. 1) there must be enough out-of-the-chain with no affiliation. 2) there must be at least one phrase containing all the terms of this popular keyword. After determining the expert page, find all the phrases on the page that contain the terms in the popular keywords or the terms 1 to 2, and divide these phrases into three levels, they are all included, with 1 difference and 2 Terms of difference, respectively for this three-level CalculationLevel score, The level is divided into allPhrase scoreThe score of a phrase is determined by the position of the phrase on the page. The score ranges from high to low, then, the expert score is obtained for the comprehensive calculation of the three levels. The following is a simple example of the popular keyword "automobile consumption". The homepage and link page of "China automobile consumption network" are the expert pages of this keyword, because he has enough domain names and does not belong to the 315che.com host domain names and the same class C ip's outbound chain, the "China automotive consumer network" in the title also contains the terms "Automobile" and "consumption. Next, calculate the score of the first level (including the phrases in all terms). The phrase "China automotive consumer network" scored 16 points in the title (hypothesis ), in the anchor text, the "China's automobile consumption and Financial Management trend survey" scored 1 point. Then, the first grade score is 17 points, and the second grade is calculated (a difference in terminology ), level 3 (two terms worse ). In this way, the weighted sum of the three levels isExpert score,Note that the weights of these three levels vary greatly. the weights of the original algorithms from level 1 to level 3 are 2 ^ 32, 2 ^ 16, and 1, because HillTop prefers exact matching.

II, Rating the target page. The score on the target page on an expert page is equal to the score on the expert page. × Number of phrases that can be differentiated on the expert page..Take the first N expert pages pointing to the target page. For Multiple expert pages with the same membership, point to the target page, and take the expert pages with the highest score, then the sum of the scores of these expert pages on the target webpage is obtained. This page corresponds to the score of this popular keyword, which is calledIndustry score.

We can see that the HillTop algorithm ensures the relevance of the evaluation results to keywords through different levels of scoring, and ensures the relevance of the subject (industry) through scoring at different positions, the number of differentiated phrases prevents keyword stacking.

 

Summary:The HillTop algorithm has a game idea. In terms of links, websites in the same industry need both competition and cooperation, only websites that are "recognized" by their peers can query top key keywords. HillTop basically destroys the hope of small websites for popular keywords. Unless you have strong expectations for popular keywords, this traffic will only last for a short time. Of course, HillTop is only an important factor in ranking, not all.