OllamaQVector - Vector Search API

Overview

This project provides a vector search API built with LangChain, Qdrant, and Ollama for embedding-based retrieval. The API allows users to create collections, add documents, and perform similarity searches using vector embeddings.

Setup and Installation

Prerequisites

Ensure you have the following installed:

• Docker
• Docker Compose

Running the Project

1. Clone this repository:

   git clone [<repository-url>](https://github.com/ana3A/OllamaQVector.git)
   cd OllamaQVector

2. Start the services using Docker Compose:

   docker-compose up -d --build

   This will initialize the following services:

   • App (Express server running the API)
   • Qdrant (Vector database for storage and retrieval)
   • Ollama (Embeddings model server)

3. Pull the embedding model: After the containers are running, you need to pull the required embedding model. Run:

   docker exec -it ollama ollama pull nomic-embed-text

   If using a different Ollama embedding model, update embeddings.service.ts with the correct model name and pull it using the same command.

API Endpoints

Collections

• Create a collection: POST /collections/create
• List collections: GET /collections/list
• Delete a collection: DELETE /collections/:collectionName
• Get collection metadata: GET /collections/:collectionName/metadata

Documents

• Add a document: POST /documents/add
• Add multiple documents: POST /documents/add-multiple
• Delete a document: DELETE /documents/:collectionName/:documentId
• Update a document: PUT /documents/update

Search

• Perform similarity search: POST /search/search

Environment Variables

Variable	Default Value	Description
PORT	3000	API server port
QDRANT_HOST	http://qdrant:6333	Qdrant service URL
OLLAMA_URL	http://ollama:11434	Ollama service URL

Project Structure

├── controllers/
├── routes/
│   ├── collection.router.ts
│   ├── document.router.ts
│   ├── search.router.ts
├── services/
│   ├── document.service.ts
│   ├── embeddings.service.ts
│   ├── qdrant.service.ts
│   ├── vectorStore.service.ts
├── Dockerfile
├── docker-compose.yaml
└── README.md

Stopping the Services

To stop the running containers, execute:

docker-compose down

Notes

• The ollama pull command must be run after the container is initialized for the first time.
• If using a different embedding model, ensure you modify the EmbeddingService class in embeddings.service.ts to reflect the correct model name.

License

MIT License