Core Concepts Wiki Entry

BERT

Bidirectional Encoder Representations from Transformers - a language model used in search.

BERT, deployed in Google Search in 2019, helps the algorithm understand the context and intent of conversational queries by analysing words bidirectionally rather than sequentially. BERT particularly improved interpretation of prepositions, negations and ambiguous phrasings. The technology forms a foundation for subsequent language models including MUM and Gemini.

Related Entries

RankBrain

A machine learning system Google uses to interpret search queries.

MUM

Multitask Unified Model - Google's multimodal AI for complex search queries.

Search Algorithm

The complex set of rules and machine learning systems that rank web pages.

Want to apply this in practice?

The Complete Wix SEO Course covers this topic and 67 others in step-by-step lessons designed for real Wix sites.

Explore the course