KeyBERT for SEOs: Google Colab Template

This ready-to-run Google Colab notebook harnesses KeyBERT’s BERT-powered keyword extraction to automatically identify the core themes within your SEO keyword lists. Created by Lazarina Stoy for MLforSEO.com, this template transforms messy keyword data into organized clusters by extracting the most semantically relevant 1-gram and 2-gram phrases from each keyword, making it easier to group related terms and understand search intent patterns across your keyword portfolio.
The workflow is straightforward: upload a CSV or Excel file containing a “Keywords” column, and the notebook processes each keyword to extract its core single-word term (1-gram) and its most relevant two-word phrase (2-gram). KeyBERT uses pre-trained BERT embeddings to understand semantic similarity, so it identifies the phrases that best represent each keyword’s meaning—not just the most frequent words. The processed results include your original keywords plus two new columns showing the extracted core terms, which you can then use to create semantic keyword clusters, identify topic patterns, or organize content strategies. The output automatically downloads as a CSV file ready for further analysis.
Use this for:
‧ Automatically clustering hundreds or thousands of keywords into semantic topic groups based on their core themes
‧ Identifying the main search intent behind keyword variations by extracting their most meaningful phrases
‧ Organizing keyword research data for content planning and information architecture decisions
‧ Finding natural groupings in keyword lists to inform content hub strategies or site structure
‧ Simplifying keyword analysis by reducing long-tail keywords to their essential semantic components
This is perfect for SEO professionals and content strategists who need to make sense of large keyword lists quickly, identify natural topic clusters, and organize keywords by semantic meaning rather than manual sorting or simple pattern matching.