Product categorization with machine learning

Why product categorization

  • physical stores have labels -> so should online stores
  • easier search for users - better discoverability
  • higher conversions for online stores
  • categories on webpages are better signals for search engine algorithms -> better SERPs
  • pages aggregating products in category -> more webpages for indexing by search engines -> more visitors

Taxonomies

  • For general texts -> IAB taxonomy
  • For e-commerce texts -> Google Taxonomy (3+ Tiers, 1000+ categories)
  • Another taxonomy for e-commerce texts, like products is Facebook taxonomy

Machine learning models for product categorization

  • Support Vector Machines
  • Logistic Regression
  • Recurrent Neural Nets
  • Transformer models

Vectorization of texts (needed for ML)

  • Bag of Words (BOW)
  • TF-IDF
  • Word2Vec
  • Glove
  • Sentence embeddings
  • BERT

Free tools for product categorization

deck

By categorization

deck

  • 129