Vectors

Vectors are used to enable semantic search within Aiimi Insight Engine. Items within your system are vectorised during a Python Rest Enrichment Step and grouped together based on the vectors created in mappings.

A Vector must be created before the Python Rest enrichment step.

Create a New Vector

Vectors cannot be deleted from the Control Hub. If a vector needs removing speak to your Elastic administrator.

1. Within the Control Hub go to Mappings > Vectors.

2. Select Create New Vector.

3. ID - Enter a unique ID for this vector.

   • This must be in camel case and only contain alphanumeric characters and underscores.

   • The Vector ID must match the enrichment steps Model ID.

4. Vector Name - Enter a user friendly name for this vector.

5. Description - Give the vector a description to explain what it is for within Description.

6. Type - At the moment we are only able to support Dense Vector types.

7. Element Type - Select the structure of your vectors from the dropdown.

   • Choose between Float or Byte.

8. Dimensions - Enter the number of Dimensions to compare for similarity.

   • Indexed vector dimensions must be less than 1024.

   • Non-Indexed vector dimensions must be less than 2048.

9. Similarity Metric - Select a metric to use to compare your vectors from the dropdown.

10. Count - Enter the number of vectors to create.

11. Index? - Check this to allow this field to be searched using the kNN search API.

12. Select Create Vector to finalise.