How to compute the cosine similarity between different subcategories in one column in pandas/sklearn? (Knowledge based recommendation)

Question

I\'m trying to create a program that recommends skincare products based on user input.

Desired result:

I would like to iterate through each su