Hey guys,
I have tried to use the openai/text-embedding-3-small in Vectorshift, but it doesn't support csv integration (only if you process the files into text).
I have tried using several types of documents such as pdfs, docx and even process the csv files into text, but I am not having good outputs.
I am trying to use it as a property recommendation system for Real Estate Agencies, the properties are structured on the document in this way:
Type: Apartament/House/...
Address: Street Located
Neighborhood:
Bedrooms:
Bathrooms:
...
The query sent to Vectorshift is structured in the same way as the properties on the documents.
Does any of you know if there is a better way to structure the query and the properties? So I can have more precise outputs.