063. BERT – What we know (Tips)

063. BERT – What we know (Tips) 780 522 Create 8

Does Google’s new algorithm make much of a difference?

Google is making significant changes to the way we find what we are looking for, and BERT has initially been heralded the most prominent change they have made to processing in over five years. But here at Create8, we know what you’re thinking. What exactly is BERT, and why has it been named after one half of the Sesame Street duo?

BERT, which stands for Bidirectional Encoder Representations from Transformers, is a system by which Google can attribute meaning to any ill-phrased sentences, therefore only displaying searches that are relevant to the language used in your initial search. For example, throughout the English language, there are many words that have more than one meaning, searching them left Google choosing carefully between the information it returned. Before BERT, Google did a very poor job of this due to the inability to attribute context to searches. For example, the term ‘evening’ used in a poorly worded sentence, would previously have returned relevant searches for the night or the smoothing out of something. Now Bert takes the context of the words surrounding the homograph into account.

SEO and BERT

We understand your struggle when it comes to SEO. The time-consuming mind-numbing process of tailoring your site to your audience is something that very few get a kick out of, web designers included. However, when it comes to the BERT implementation, the benefits will be in favour of the well-worded and concise websites. Removing all of those ‘what are you talking about’ moments when trawling through Google.

What next?

Many experts would highlight that BERT is only the start when it comes to computer recognised linguistics, but in its very nature, it is a huge leap forward. Learning to differentiate when specific web pages are light of hard copy information, will be a difficulty, and ranking the value of a page to a searcher will need some work. However, when it comes to placing recognition on specific lingual devices, it seems computers have well and truly taken their place at the front of the race.