It would be great if a LLM would "tag" autimatically queries based on their semantics or meaning. As a searchmanager i would like to train the model based on a few examples and define the classification the model should use based on my needs. Then the machine would follow those rules and automatically do the job for me on the whole existing and incoming data.
Queries for e.g. :
cigaret > Food
T-Shirt > Nonfood
Or:
Nike > Brandsearch
E-bike > Productsearch
Fischer Cita ECU 1401 > Spedific product search
As you can see based on this examples... it would be great to be able to customize those attributes or categories because its very assortment related.