Find the nearest neighbors to a provided search item

This API call allows you to query an entity for results which are the closest match to an image or text you provide to Graft through "entity_knn_adhoc" this can be a string or uri to an image of interest.

For entity_knn_adhoc, this can be one or all of the fields embedded within your embedding.

For example if your Entity has an embedding (default name is my_embedding) using a single field called "description" the syntax is...

"entity_knn_adhoc": {"description":"The most delicious food"},

If your embedding is made up of multiple fields you can search on either a single field or all of the fields contained in the embedding.

For example if your Entity has an embedding using the fields "description", "cuisine" and "customer_rating" you can provide search parameters for any of those fields or all

"entity_knn_adhoc": {"description":"The most delicious food"},
or
"entity_knn_adhoc": {"cuisine":"modern italian"},
or
"entity_knn_adhoc": {"customer_rating":"above average"},
or
"entity_knn_adhoc": {"description":"The most delicious food","cuisine":"modern italian","customer_rating":"above average"},

Note: You can not provide 2 fields in this example it MUST be One or All

Sample Code

curl https://app.graft.com/v1/query -H 'Content-Type: application/json' -X POST -d '{
  "secret": "SArh7qq.4q9vhmwgh17JAJ6EuhAlJJUYavhtXVgiRTabcmgS",
  "organization": "<your_organization>",
  "project": "my_project",
  "output_format": "csv",
  "entity": "MyEntity",
  "embedding_name": "my_embedding",
  "entity_columns": ["id", "data"],
  "entity_knn_adhoc": {"<field to be searched>":"The most delicious food"},
  "entity_knn_k": 10
}'
 
 

Simple Example

Please see REST API overview for the example values

"Find the 5 nearest matches to the provided speciality"

curl https://app.graft.com/v1/query -H 'Content-Type: application/json' -X POST -d '{
  "secret": "somethingiamnotgoingtosaveingithub",
  "organization": "supervillain",
  "project": "world_domination",
  "output_format": "csv",
  "entity": "despicable_plan_1",
  "embedding_name": "my_embedding",
  "entity_columns": ["id", "name", "speciality", "number_of_films", "id_photo"],
  "entity_knn_adhoc": {"minion_speciality":"various nefarious skills including stealth"},
  "entity_knn_k": 5
}'
 