One of the basis of search engine ranking is the correlation between keywords and web pages.
The difference between machine algorithms and humans is that humans can directly understand the meaning of words and articles, but machines and algorithms cannot. When people see the word apple, they know it refers to a round, watery and delicious thing, but search engines cannot understand it perceptually.
But search engines can grasp the relationship between words, which involves semantic analysis.
Two years ago, the search engine optimization industry was talking about latent semantic indexing (Latent Semantic Indexing) very enthusiastically. Because it’s been a long time, I don’t remember it very clearly, so I’ll give you a rough introduction.
The so-called latent semantic index refers to how to find the relationship between words through massive documents. When two words or a group of words appear in large numbers in the same document, these words can be considered to be semantically related.
For example, the words computer and computer are often used interchangeably when people write articles. If these two words appear in a large number of web pages at the same time, search engines will consider these two words to be extremely semantic. Relevant.
It should be noted that latent semantic indexing does not depend on language, so although SEO and search engine optimization are one in English and one in Chinese, these two words appear in large numbers in the same web pages. Although Search engines do not yet know what search engine optimization or SEO refers to, but they can semantically connect words such as "SEO", "search engine optimization", "search engine optimization" and "SEM" together.
Another example is the words apple and orange, which also appear in large numbers in the same documents, but the closeness is lower than that of synonyms.
Whether search engines use latent semantic indexing has not yet been determined, because search engines neither admit nor deny it. But one fact is that in 2002, Google bought Allied Semantic, a company with patents for latent semantic indexing.
This semantic analysis technology can give us some tips on search engine optimization.
The formation of website themes
Usually a website with proper logic and structure will be divided into different channels or columns. Talk about somewhat different but closely related topics in different channels. These topics together form the theme of the website. After the search engine includes the pages of the entire website, it can determine the theme of the website based on the semantic correlation between these subject words.
Web content writing
Since two years ago, there has been a phenomenon in search engine rankings. When searching for a certain keyword, the web pages that rank at the top sometimes do not even contain the search term. Keywords, this is most likely latent semantic indexing at work.
For example, if you search for computers, one of the top web pages may contain an article that only mentions computers but does not mention computers. Because search engines know through semantic analysis that these two words are closely related.
Another thing to note is that when writing web pages, you should not be limited to the target keywords. You should include words that are semantically related to the main keywords to support the main keywords.
This is also reflected in the search results. Although some articles contain a large number of main keywords, they lack other supporting vocabulary and often do not rank well.