Google's search engine is one of the most popular and widely used tools for finding information and answers to various questions. The search engine is utilized worldwide, providing users with the ability to quickly locate the necessary data. This is why the Google development team is continuously working to improve its search system.
Currently, every query is perceived by Google's search engine as a set of words, for which relevant results are matched. The system struggles with conversational and complex queries, and understanding language remains a persistent challenge over time.
In the near future, the company plans to implement a new natural language processing algorithm, based on the BERT (Bidirectional Encoder Representations from Transformers) neural network introduced last year. The algorithm can analyze a query in its entirety, without breaking it into words and taking into account prepositions and conjunctions. This approach will provide a complete context for the query, yielding more suitable answers.
Google's developers state that creating the algorithm based on the BERT neural network is "the most important achievement in the last 5 years and one of the most significant events in the entire history of the search engine." Currently, the new algorithm is used to process a portion of queries to Google's search engine performed in English. In the future, the algorithm will be extended to all supported languages, but it remains uncertain when this will occur.
Source: 3dnews.ru
