Classify and Label Keywords

By training a customised Machine Learning (ML) Model, new unseen keywords can be automatically labeled (classified) into all kind of categories (e.g: Brand, Brand Compound / Navigational, Competitors or even sentimental). Supporting misspelling and all language sign's in utf-8.

Insights & Advantages

• No need of Regular Expressions (captures all brand misspelling's)

• Multi Label Classification (100+ Categories) e.g: Brand, Brand +Support, Navigational, Brand Negativ, Brand Positive, Multilingual Twin's)

• Learning semantical relations within your keyword spectrum (n-grams / short-, longtail-keywords)

• Use the new predicted categories to forecast impressions & seasonality 

• Supporting all utf-8 language sign's / characters 

• Re-train the Model with new data (manual 'reinforced learning') to make it more and more precise over time (99,9%)

Example Use Cases

• Classify Google Search Console (GSC) Keywords and compare daily searches vs. average.

• Classify & Compare GSC vs. Adwords Keyword Cluster

• BigQuery ML / API (Experimental). Include Keyword Clustering / Classification into your daily GSC Export pipeline.

• Use SimpleML for Sheets to make forecast's by keyword cluster (category) 

• Visualise semantical similarities (multilingual) with keyword embeddings & Tensorflow Projector 

• You have your own case?

Interested?

1. Try the open-source playground model at Github (Upcoming)
   (Important: Depending on the input data, the model needs a lot of data pre-processing, analyses, hyperparameter- and deep neural network tuning to archive a good accuracy. Always test the live model on Responsible AI QS.

2. Maybe I can support you in optimising your own customised model? From creating a Data Card till testing the model on Responsible AI Quality Standards and Requirements, feel free to contact me.

3. Integrate / create the ML Model directly inside your BigQuery Pipeline (Experimental only).