If you’ve already worked with ChromaDB, getting started with Weaviate is quite straightforward.

Both are vector databases that can store objects, embeddings, and metadata, then retrieve data using semantic search.

What is Weaviate?

Weaviate is an open-source vector database designed for AI applications.

Just like ChromaDB, you can use it to:

• Store objects and embeddings
• Perform semantic search
• Filter data
• Build RAG applications
• Build recommendation systems

One of the nice things about Weaviate is that it also provides keyword search and hybrid search out of the box.

Weaviate vs ChromaDB

Since we already know ChromaDB, here’s the simple comparison:

In short:

ChromaDB
→ Simple and easy to get started

Weaviate
→ More built-in search capabilities

For a simple local project, ChromaDB can be enough. If you need more advanced search capabilities, Weaviate is worth considering.

Using Weaviate

Let’s use a movie dataset as an example.

1. Install

pip install weaviate-client pandas

2. Create a collection

Weaviate organizes data into collections, similar to collections in ChromaDB.

import weaviate
from weaviate.classes.config import Configure, Property, DataType

client = weaviate.connect_to_local()

movies = client.collections.create(
    name="Movie",
    vector_config=Configure.Vectors.text2vec_sentence_transformers(),
    properties=[
        Property(name="title", data_type=DataType.TEXT),
        Property(name="description", data_type=DataType.TEXT),
        Property(name="genres", data_type=DataType.TEXT_ARRAY),
    ],
)

3. Insert data

movies.data.insert(
    properties={
        "title": "Extraction",
        "description": "A black-market mercenary...",
        "genres": ["Action", "Thriller"],
    }
)

For multiple objects, we can use batch insertion:

with movies.batch.dynamic() as batch:
    for movie in movie_data:
        batch.add_object(properties=movie)

4. Semantic search

We can search by meaning using near_text():

response = movies.query.near_text(
    query="funny movies for children",
    limit=5,
)

for result in response.objects:
    print(result.properties)

Instead of looking for the exact words, Weaviate searches for movies that are semantically related to the query.

5. Hybrid search

We can also combine keyword and semantic search:

response = movies.query.hybrid(
    query="thriller",
    alpha=0.5,
    limit=5,
)

Here:

alpha = 0
→ Keyword search

alpha = 0.5
→ Keyword + Vector search

alpha = 1
→ Vector search

6. Find similar objects

Weaviate can also find objects similar to an existing object:

response = movies.query.near_object(
    near_object=movie_id,
    limit=5,
)

This is useful for recommendation features such as:

Extraction
    ↓
Similar movies
    ↓
Unhinged
The Contractor
...