A common problem in large online stores is the need for a product to be present in several categories. With staff turnover, it takes a long time for new employees to study the entire category structure of hundreds of categories. The lack of a product in the right categories reduces the likelihood that it will be noticed by the consumer and purchased. My client's site has over 450 categories and often employees do not add products where they should or do not add them in enough categories. That's why I developed an automatic categorization of products for his online store with the help of machine learning. The system is self-learning, training a model loading the descriptions of all products of all categories and is easy to integrate. When an employee writes the product description, he is automatically offered categories in which to add it. He can remove one or add more. The categories have a written probability of coincidence - the higher the higher the categorization. If you are interested, I can send a link to test the automatic categorization.