Google New Algorithm

SMITH Google new algorithm

SMITH (SIAMESE MULTI-DEPTH TRANSFORMER-BASED HIERARCHICAL ENCODER) is a new model search engine algorithm by Google, Which is trying to understand passages, queries, and documents. If you are a webmaster then you must have to know about SMITH algorithm, it will have a big impact on SEO. Content is the king for SEO (Search Engine Optimization). Cracking content is neither easy for SEO nor for Google. The Google designed SMITH model to understand long paragraphs and the complete document. SMITH is better than BERT algorithm because it has the ability to understand content whereas BERT understands words and sentences.

What is BERT?

BERT (Bidirectional Encoder Representations from Transformers) launched by Google on October 25, 2019. BERT was the first algorithm to leverage bidirectional processing to explain the meaning of a single word based on the contextual cues around it. BERT is the natural language processing algorithm. It is using in 70 languages globally. BERT is bidirectional because it can read in two directions as like left to right or right to left both, but other languages model can either read left to right or right to left it means they are
unidirectional. It helps in both languages English and Hindi. It works best in short text, not long-form.

SMITH can/ BERT can’t

SMITH can understand paragraphs that are based on documents. It is good at making long-to- long semantic connections.
SMITH is the instant pot – you need to put ingredients and it will prepare the whole meal in just a few minutes. There’s another benefit of SMITH algorithm that it helps with long-tail queries. It will be able to check the quality of the content.

BERT can understand words that are based on the passage. It is useful for short-to-long or short-to-short semantic matching. It helps with short queries.

.
How SMITH works?

SMITH works on two tower structure. One tower is the sentence level and the other one is the document level. BERT and SMITH are pre-trained models. Both can train and fine tune themselves without any supervision. SMITH can’t be a replacement of BERT. They will work
together to understand document content. This algorithm tries to understand big paragraphs, passages, and relationships of sentences and extract their meanings. According to research paper, SMITH can process 400% more text from BERT. It can understand the long text well. Google has not confirmed that they are using the SMITH algorithm. However, the research papers published that the SMITH algorithm can understand long-form queries and content. Until we get confirmation from Google we can’t say that how it will be applied.