AryaBot: News Research Tool

AryaBot is a user-friendly news research tool designed for effortless information retrieval. Users can input article URLs and ask questions to receive relevant insights from the stock market and financial domain.

Features

• Load URLs or upload text files containing URLs to fetch article content.
• Process article content through LangChain's UnstructuredURL Loader
• Construct an embedding vector using OpenAI's embeddings and leverage FAISS, a powerful similarity search library, to enable swift and effective retrieval of relevant information
• Interact with the LLM's (Chatgpt) by inputting queries and receiving answers along with source URLs.

Installation

1.Clone this repository to your local machine using:

  git clone https://github.com/codebasics/langchain.git

2.Navigate to the project directory:

  cd 2_news_research_tool_project

3. Install the required dependencies using pip:

  pip install -r requirements.txt

4.Set up your OpenAI API key by creating a .env file in the project root and adding your API

  OPENAI_API_KEY=your_api_key_here

Usage/Examples

1. Run the Streamlit app by executing:

streamlit run main.py

2.The web app will open in your browser.

• On the sidebar, you can input URLs directly.

• Initiate the data loading and processing by clicking "Process URLs."

• Observe the system as it performs text splitting, generates embedding vectors, and efficiently indexes them using FAISS.

• The embeddings will be stored and indexed using FAISS, enhancing retrieval speed.

• The FAISS index will be saved in a local file path in pickle format for future use.

• One can now ask a question and get the answer based on those news articles

• In video tutorial, we used following news articles

  • https://www.moneycontrol.com/news/business/tata-motors-mahindra-gain-certificates-for-production-linked-payouts-11281691.html
  • https://www.moneycontrol.com/news/business/tata-motors-launches-punch-icng-price-starts-at-rs-7-1-lakh-11098751.html
  • https://www.moneycontrol.com/news/business/stocks/buy-tata-motors-target-of-rs-743-kr-choksey-11080811.html

Project Structure

• main.py: The main Streamlit application script.
• requirements.txt: A list of required Python packages for the project.
• faiss_store_openai.pkl: A pickle file to store the FAISS index.
• .env: Configuration file for storing your OpenAI API key.