Say I have a set of marketplace listings in natural language, with a possible image. These are labeled with the product type, quantity, ppu, etc. Is there an algorithm that performs well in learning the labels of the product in the listing when given plain text? Ideally, the solution would go beyond regex, and be more adaptive. The output fields are always the same, and the input can vary in writing style and length.
We are offering bread! It is the best bread, so fluffy, with a hint of vanilla. We have 100 loaves readily available for shipping and ask $1.20 per loaf. Buy now and we'll ship before it goes stale!
type: baked goods; stock: 100; ppu: $1.20