Running Vector Query with Filter on Weaviate v4

Support
1

I have generated vector embeddings using the AWS Titan Embedding model and uploaded them with their properties into Weaviate. I’m trying to run queries on specific programs. I have question about the best approach in v4 version. Below is the code I used:

Python

# Step 2: Define the collection for the 'Document' class
from weaviate.classes.config import Configure, Property, DataType

collection_snap = {
    "class": "Document_snap",
    "description": "A class to represent documents of SNAP",
    "vectorizer": "none",  # Set to "none" because embeddings are provided
    "moduleConfig": {
        # "text2vec-openai": {},  # Configure if using OpenAI vectorization
        # "generative-openai": {}  # Configure if using generative queries
    },
    "properties": [
        {"name": "url", "dataType": ["text"]},
        {"name": "title", "dataType": ["text"]},
        {"name": "chunks", "dataType": ["text"]},
        {"name": "program", "dataType": ["text"]},  # includes 'program1', 'program2', 'program3'
    ]
}

# Batch Upload:
with client.batch.dynamic() as batch:
    for i, row in df_combined.iterrows():
        print(f"Importing document: {i+1}")
        properties = {
            "url": row["url"],
            "title": row["title"],
            "chunks": row["chunks"],
            "program": row["program"],  # includes 'program1', 'program2', 'program3'
        }
        batch.add_object(
            collection="Document_snap",  # Specify the collection name here
            vector=row["embeddings"],
            properties=properties
        )
    failed_objs_a = client.batch.failed_objects
print('batch import successful')
print(len(failed_objs_a))

# Filter programs based on user choice:
User_request_program_options = {
    'p1': 'program1',
    'p2': 'program2',
    'p3': 'program3',
    'All': ['program1', 'program2', 'program3']
}

Question 1: Is this the right way to run a query with a filter on a single program?

Python

# Query 1:
User_program_request = User_request_program_options['p1'] #program1

document_collection = client.collections.get("Document_snap")
response = document_collection.query.near_vector(
    filters=(
        Filter.by_property("program").equal(User_program_request)
    ),
    near_vector=query_vector, 
    limit=2,
    return_metadata=MetadataQuery(distance=True)
)

for o in response.objects:
    print(o.properties)
    print(o.metadata.distance)

query = collection.query.fetch_objects(
    filters=(
        Filter.by_property("program").equal(User_program_request)
    )
)

Title: Running Vector Query with Filter on Weaviate v4 - Multiple Programs Filter

Question 2: How about if the user’s choice is the ‘All’ option, which includes all three programs? Is this the right way to run the query?

Python

Query 2:

User_program_request = User_request_program_options[‘All’] # ‘program1’, ‘program2’, ‘program3’

document_collection = client.collections.get(“Document_snap”)
response = document_collection.query.near_vector(
filters=(
Filter.by_property(“program”).equal(User_program_request)
),
near_vector=query_vector,
limit=2,
return_metadata=MetadataQuery(distance=True)
)

for o in response.objects:
print(o.properties)
print(o.metadata.distance)

2

Is there anyone that can help me with Vector query with filter? I appreciate any help!

3

Hi!

You can check if the user has selected All and never apply that filter.

something like this:

#base filter
filters=(
    Filter.by_property("color").equal("grau") & 
    Filter.by_property("price").less_or_equal(300)
)

if selected != "All" or selected != None:
    filters = filters & Filter.by_property("program").equal(selected)

query = collection.query.fetch_objects(
    filters=filters
)

for object in query.objects:
    print(object.properties)

We were discussing more on conditional filters in code here:

Let me know if this helps!

4

That helped, thank you!

1 Like