|
Google and NLP: natural language processing integrated into the search engine algorithm In terms of automatic natural language processing, Google is a reference, but we will primarily focus on how this technology is used to transform the indexing and positioning processes of web pages. To understand how the Google algorithm evolves, we must always look at the user experience. The Mountain View firm wants to guarantee the satisfaction of Internet users who use its search engine by offering them results that are as relevant as possible, which requires continuously improving the quality of the pages highlighted in its SERP.
In this context, understanding the requests made by users is a major issue. It is no longer just a matter of grasping the overall meaning of the words, but of identifying the intention behind the search in order to better respond to it. To do this, you must BTB Directory understand the nuances of a query, but also detect the terms that express a “feeling”. This work by Google on NLP led to the launch, in , of the BERT algorithm – the most significant update in five years for the firm in its own words and a real leap forward in operation search engines.
Because BERT no longer simply processes queries word for word: it weaves links between the terms used in order to take into account the context of the search and grasp the “deep meaning”. With this in mind, it looks at all the terms used, including linking words and prepositions, and evaluates the “sentiment” that emerges from the query by assigning it a score positive, negative or neutral. At the time of its launch, the BERT algorithm for Bidirectional Encoder Representations from Transformerswas the technological culmination of Google's research in NLP.
| 
發表於 14:15:02