Examples
Semantic Search with Embeddings
Store and search using vector embeddings for semantic similarity.
CogDB supports vector embeddings for semantic search and similarity queries.
Storing Embeddings
Use put_embedding() to store vectors:
from cog.torque import Graph
g = Graph("vectors")
# Store embeddings
g.put_embedding("king", [0.5, 0.2, 0.9])
g.put_embedding("queen", [0.48, 0.21, 0.88])
g.put_embedding("apple", [0.1, 0.8, 0.3])Similarity Search
Find similar items using cosine similarity with sim():
# Find items with cosine similarity > 0.9 to "king"
g.v(["king", "queen", "apple"]).sim("king", ">", 0.9).all()
# {'result': [{'id': 'king'}, {'id': 'queen'}]}Similarity Operators
The sim() method supports these comparison operators:
| Operator | Description |
|---|---|
> | Greater than |
>= | Greater than or equal |
< | Less than |
<= | Less than or equal |
== | Equal to |
Examples
# Find very similar items (similarity > 0.95)
g.v(["king", "queen", "apple"]).sim("king", ">", 0.95).all()
# Find somewhat similar items
g.v(["king", "queen", "apple"]).sim("king", ">", 0.5).all()
# Find dissimilar items
g.v(["king", "queen", "apple"]).sim("king", "<", 0.5).all()
# {'result': [{'id': 'apple'}]}