The other kind of “semantic search”

When we think “semantic search” we think embeddings.

This is just one kind of “semantic search”. There’s a less sexy approach: mapping queries + documents into hierarchies.

A query for leather couch maps to: /Furniture / Living Room / Seating / Couches and Sofas

Then we rank products based on proximity to the query’s hierarchy

1. Item: Loveseat mapped to /Furniture / Living Room / Seating / Couches and Sofas
2. An “easy-chair” comes next - its a sibling in the hierarchy ``/Furniture / Living Room / Seating / Recliners`
3. Maybe a “living room table” would come next as a cousin - its a /Furniture / Living Room / Tables /...

Or maybe you decide at the “cousin” level to remove those entirely!

This style of semantic search puts YOU in the driver’s seat. Letting you define similarity close to your domain.

-Doug

​

Services: Training (use code search-tips) · Consulting​

This is part of Doug’s Daily Search tips - subscribe here