{"id":14386,"date":"2021-01-11T15:25:38","date_gmt":"2021-01-11T15:25:38","guid":{"rendered":"https:\/\/www.improvemysearchranking.com\/?p=14386"},"modified":"2023-12-05T11:10:31","modified_gmt":"2023-12-05T11:10:31","slug":"googles-new-smith-algorithm-and-how-it-outperforms-bert","status":"publish","type":"post","link":"https:\/\/www.improvemysearchranking.com\/googles-new-smith-algorithm-and-how-it-outperforms-bert\/","title":{"rendered":"Google\u2019s new SMITH algorithm (and how it outperforms BERT)"},"content":{"rendered":"

Google has a new search engine algorithm, SMITH. And according to Google, it is outperforming Google BERT in understanding long-form queries and content.<\/span><\/p>\n

It remains a mystery whether or not Google is using the SMITH algorithm. It is important to note that Google rarely says which specific algorithms it is using at a given time. Therefore, Google may or may not be using SMITH as of now.<\/span><\/p>\n

That, however, does not diminish its value and the need for understanding how this algorithm works. In my opinion, this gives a fascinating insight into the direction Google is moving as a search engine and how it sees the future of online content and content consumption.<\/span><\/p>\n

<\/p>\n

What is SMITH?<\/b><\/h2>\n

Put simply, SMITH or Siamese Multidepth Transformer-based Hierarchical Encoder is a new search engine algorithm by Google that focuses on understanding long-form documents. More specifically, SMITH is particularly good in understanding the context of certain passages within long-form content.<\/span><\/p>\n

How is SMITH different from BERT?<\/b><\/h2>\n

SMITH and BERT appear to be related, and SMITH seems like an extension of BERT.<\/span><\/p>\n

While SMITH deals with understanding passages within the context of documents, BERT is trained to understand words within the context of sentences.<\/span><\/p>\n

When it comes to understanding long-form content, BERT has limitations that SMITH does not.<\/span><\/p>\n

According to a research whitepaper by Google:<\/span><\/p>\n

\u201cIn recent years, self-attention based models like Transformers\u2026 and BERT \u2026have achieved state-of-the-art performance in the task of text matching. These models, however, are still limited to short text like a few sentences or one paragraph due to the quadratic computational complexity of self-attention with respect to input text length.<\/span><\/i><\/p>\n

In this paper, we address the issue by proposing the Siamese Multi-depth Transformer-based Hierarchical (SMITH) Encoder for long-form document matching. Our model contains several innovations to adapt self-attention models for longer text input.\u201d<\/span><\/i><\/p>\n

The whitepaper also explains why understanding long documents could be more difficult:<\/span><\/p>\n

\u201cSemantic matching between long texts is a more challenging task due to a few reasons:<\/span><\/i><\/p>\n