What is BERT?
BERT is a deep-learning model developed by Google, based on ‘the transformer architecture’.
It’s a content-related update, aimed at improving the way Google interprets the context and search intent of queries, to better understand of our relationship with words.
Google’s BERT Announcement
In 2018, Google introduced (and open-sourced) a “neural network-based technique for natural language processing (NLP) pre-training, called Bidirectional Encoder Representations from Transformers” or BERT for short.
It follows Google’s research on transformers: “models that process words in relation to all the other words in a sentence, rather than one-by-one in order. BERT models can therefore consider the full context of a word by looking at the words that come before and after it—particularly useful for understanding the intent behind search queries.” – Google AI Blog
In summary, BERT models help Google to better understand the context and intent behind users’ search queries.
Natural Language Processing (NLP)
Natural language processing is a branch of AI dealing with linguistics. Advances in NLP help machines to understand how humans naturally communicate.
Transformer refers to a model that uses ‘attention’ mechanisms to improve the quality of machine translation systems, and boost the speed with which they can be trained.
‘Attention’ is a mechanism which allows the model to focus on the relevant parts of the input sequence, as needed. These mechanisms highly improved the quality of machine translation systems, while being more parallelizable and requiring significantly less time to train.
The Scope of BERT
Google have said the BERT update impacts 10% of search queries. More specifically:
- BERT is currently used for general ranking purposes only in US English. No timeline has been provided yet for when BERT’s scope might broaden out to include other languages.
- However, BERT is functional for featured snippets for all languages in which featured snippets are available (currently there are approximately 25 supported languages).
Examples of BERT in action
Here are some other examples where BERT has helped Google grasp the subtle nuances of language that computers don’t quite understand the way humans do. BERT has managed to identify the important contextual distinction added by ‘stopwords’ that would previously have been ignored (or at least, their significance would have been under-valued).
Optimising for BERT
BERT helps Google better understand important nuances in the way we talk (or type). As such, you can think of it more as a ‘correction’ in machine understanding than a new feature or ranking factor that you have to optimise for.
In fact, Google’s Danny Sullivan tweeted:
“There’s nothing to optimize for with BERT, nor anything for anyone to be rethinking.
The fundamentals of us seeking to reward great content remain unchanged.”
As usual, ensure that you help search engines (and users) understand your content well by providing them with a context where you can, in the form of:
- Coherent, detailed content (i.e. no ‘thin’ content)
- Structured data
- Internal linking with useful anchor text
One small potential difference is that you don’t need to ignore stopwords quite as much in your SEO activities and instead can think about queries in a slightly more natural way (at least in US English queries, for now).