: How Google knows category of the website? When speaking of backlinks as we all know the most important thing is backlink is the same "Category" (relevant) as your website. So how Google knows

@Caterina187

While I agree with what @closetnoc wrote, I suspect (and this largely my personal, anecdotal experience...and common sense) there are some other significant points:

1.) Google uses image analysis...do the images on your site (file name, alt text, image content) align with the purported topic of your site? Beyond that, are they original (or are they stock photos, stolen images, etc.)

2.) Google uses backlinks that it deems trustworthy:

I did an SEO audit for a website that had a backlink like:
insurance.arkansas.gov/regulatory/registered-insurance-vendors/vendor-123456

What category would you suspect a business with a link like that is in?

What about a website with a backlink like: well-know-porn-site.xxx/best-porn-videos

I'd guess the first site could be categorized as insurance and the second should be categorized as porn.

Vote Up Vote Down

@Alves908

First things first, stop thinking in terms of keywords. In the very early days, before Google, search engines matched keywords and used other metrics including density and proximity. The original Google research paper written by Brin and Page in 1997 railed against this idea citing the limited potential to making successful and meaning matches. And they were right. Instead, Google would use semantics to determine what results would match the search query. In the early days, the use of semantics was somewhat limited, however, it was spot on in how to implement a simple and effective semantic search and has not swayed too far from it's original incarnation.

So stop thinking in terms of using keywords and how Google matches keywords. It doesn't.

Google has never made direct keyword matches. Instead, it uses several linguistic semantic algorithms and methods such as topical analysis and PoS (parts of speech) and others to determine a good match. Case in point, RankBrain is just a larger implementation of semantics applied to the search query as well as further applying AI (artificial intelligence) machine learning between the query and the results allowing Google to create better results. It is simply an advancement of what has always been done.

The HTML code is broken down to a HTML DOM (model) and evaluated. Content is broken down into segments, a semantic concept, which can be as small as a few words and as large as a whole document, however, generally, a segment would be a content block. A content block is as small as a paragraph and as large as the entire content. Typically, it is best to think of a content block is a single paragraph to as large as all the paragraphs between header tags to include the header tag immediately proceeding the paragraphs. Each content block logically and typically contains one topic whether we realize it or not. Humans naturally organize their writing so that a paragraph is different from another and paragraphs within a larger block follow a logic flow and larger topic. Each content block between headers follow a logic flow throughout the content and is strongly signaled by the header tag. Each header tag and paragraph is analyzed individually using semantics as well as the larger content blocks and the entire document. This analysis is contained within virtual tables of metrics and matrices (metrics of metrics). This is how Google knows what the content is about.

This is also done for any link text, target URL, any link within content, and placement meaning content within a content block before the link and after the link. Remember that content blocks can be the entire content. When a link is analyzed, it is scored against content blocks to determine if any metric matches exist. Add to this, these metric scores are compared to the metrics of the target page.

Where the metrics match using all of the analysis is where relevancy is found.

Now, please know that the concept of markets is also applied to sites. For example, we know about directory sites, link sites, whois sites, and so on. While the term market is not generally used by Google anymore, applying analysis to determine market is the original concept and how Google originally applied the concept. In this analysis, Google has determined it's own markets and analysis to place sites into numerous simple categories. Any site that is placed within a market is compared to other sites within the same market. In this, all metrics, semantics and search performance, are used to rate a site within it's market. Niche sites cannot escape this. A blog is a blog regardless of the topic. An e-zine is an e-zine regardless of the topic. If, for example, your blog does not compare well within it's market, it very likely will not perform well in search regardless of what your blog is about. How a site can break out of this is to create a higher quality and more appealing and popular site.

While the concept of markets does not effect link analysis, it does explain further how Google analyzes sites and determine categories (markets) which is part of your question.

Vote Up Vote Down