Vietnames RAG powered AI chatbot for an online flower shop

Running in local

• Create an .env file and populate these ENV variables needed as required in .env.example
• for VECTOR_INDEX_NAME, create a MongoDB Atlas vectorSearch index for the DB_COLLECTION collection:

{
  "fields": [
    {
      "numDimensions": 768,
      "path": "embedding",
      "similarity": "cosine",
      "type": "vector"
    }
  ]
}

• for KEYWORD_INDEX_NAME, create a MongoDB Atlas search index for the DB_COLLECTION collection:

{
  "mappings": {
    "dynamic": false,
    "fields": {
      "title": [
        {
          "type": "string"
        }
      ]
    }
  }
}

• for SEMANTIC_CACHE_INDEX_NAME, create a MongoDB Atlas vectorSearch index for the SEMANTIC_CACHE_COLLECTION collection:

{
  "fields": [
    {
      "numDimensions": 768,
      "path": "embedding",
      "similarity": "euclidean",
      "type": "vector"
    },
    {
      "path": "llm_string",
      "type": "filter"
    }
  ]
}

• Install dependencies:

python -m venv venv
source venv/bin/activate
pip install -r requirements.txt

• Run web scraper, all_urls.json and output.json files will be generated:

python web_scraper.py

• Load documents and their embeddings into MongoDB:

python load_document.py

• Start the server, the server will run from http://localhost:5000 by default:

python serve.py

• Run streamlit app to access chatbot:

streamlit run Hello.py

• To run test suite:

python -m unittest