Skip to content

Latest commit

 

History

History
106 lines (78 loc) · 3.1 KB

README.md

File metadata and controls

106 lines (78 loc) · 3.1 KB

Langchain Reproductive Health Bot

Setup and Run the Project

Prerequisites

Steps

  1. Clone the Repository:

    git clone [email protected]:jsakeag/langchain-healthbot-new.git
cd langchain-healthbot-new

  2. Install Dependencies:

    npm install

    If needed:

    pip install -r /path/to/requirements.txt

  3. Set Up OpenAI API Key:

  • Obtain your OpenAI API key.

  • Create a .env file in the root directory and add your OpenAI key.

    REACT_APP_OPEN_AI_API_KEY=your_openai_api_key

  1. Start the Backend and Frontend Servers:

    cd backend
python3 app.py
    cd frontend
npm start

Options

  1. Q/A with our online KG: The default way option uses a neo4j knowledge graph we created ourselves. Notice that the chatbot_response import is from chatbot_KG.py in the app.py file. Here is a code snippet from our chatbot_KG.py file:
kg = Neo4jGraph(
    url="neo4j+s://4221dbdc.databases.neo4j.io",
    username = "neo4j",
    password="6wmPE5kyuqGAxoPBzZeB12Mb56gzaNpP9goIC-8qdAc",
    database = 'neo4j'
)

neo4j_vector_store = Neo4jVector.from_existing_graph(
    embedding=OpenAIEmbeddings(),
    url='neo4j+s://4221dbdc.databases.neo4j.io',
    username='neo4j',
    password='6wmPE5kyuqGAxoPBzZeB12Mb56gzaNpP9goIC-8qdAc',
    index_name='embeddings',
    node_label='Information',
    text_node_properties=['text'],
    embedding_node_property='embedding'
)

  1. ChatGPT Clone: The other way to run this project is the ChatGPT default 4o mini model. This can be selected by setting the chatbot_response import to chatbot.py in the app.py file. Here is a code snippet from our chatbot.py file:

    # Initialize OpenAI Chat model
llm = ChatOpenAI(
 model_name="gpt-4o-mini",
 temperature=0.7,
 max_tokens=500,
 api_key=openai_api_key  # Explicitly pass the API key here
)
# Initialize memory for conversation
memory = ConversationBufferMemory(memory_key="chat_history", return_messages=True)

# Create embedding model
embeddings = OpenAIEmbeddings(api_key=openai_api_key)  # Also pass the API key here

# Initialize Chroma DB (vector database)
vector_db = Chroma(embedding_function=embeddings, collection_name="my_collection",persist_directory="./my_chroma_db")
  • If you are experiencing CORS errors Make sure you allow your IP to access the endpoint (look at your AWS lambda scripts)

Accessing Our Resources

Neo4j Knowledge Graph

Here is a visual link to our neo4j knowledge graph:

https://workspace-preview.neo4j.io/workspace/query?ntid=google-oauth2%7C114513607131593483170

Web Scraper

Here is a [link](url to the code we used to downlaod and scrape articles neo4j: https://colab.research.google.com/drive/1ubkOjOHJJ5TnD5pk7p4rXUMvesZP2CBW