Potential semantic index (LSI) of Seo Google algorithm parsing Series)

As a seoer, we must have a certain understanding of the search engine ranking algorithm before we can really talk about optimization. Next we will focus on a seriesAlgorithmI would like to share my understanding and experiences with you. I hope you can give me more advice and learn from each other. Here we start with the keywordCorrelation Algorithm TF/IDFStart.

The fastest way to search engine cheating is keyword stacking, which is due to the defects of relevance algorithms in information retrieval. To combat this cheating method, the search engine uses the latent semantic index (Latent Semantic Indexing, LSI) algorithms are used to discover these cheating pages. LSI is also an old algorithm in the information retrieval field. In 1988, it was developed by s.t. dumais and others proposed that it is mainly used for natural language understanding, semantic analysis of documents through statistical methods, and exploration of synonyms and related phrases. For example, the word "automobile consumption" is used to analyze a large number of pages to find that this word frequently appears in "Automobile Consumption Loan ", among the phrase "China automotive consumer network, then machines can think that people's language habits are to combine "automobile consumption" with "Automobile Consumption Loan" and "china automobile consumption network" to describe some things. Through such analysis, we can find that some keywords generated by machines pile up pages, because the search engine does not think that these related phrases will appear on the pages generated by machines.

The LSI algorithm is used in many Google applications, such as adwords, Google suggest, and anti-cheating mentioned above.

The LSI algorithm reminds us to pay attention to the keyword density of the page and the use of related phrases when optimizing the search engine, and try to use natural language methods to improve the relevance of the page.