/financial-analytic-chat-app

Primary LanguageTypeScript

Financial Analytic Chat app - Ngo Gia Duc

Email: duc.ngogia2002@gmail.com

Link Demo:

I also built CI/CD pipeline for this application. I deployed the application on AWS ECS. Detail in _infrastructure folder

Try it now: Demo

Sample Image:

chat_app_workflow.png

Relevant Project:

Check out this chat app that I made few months ago! It uses a language model to summarize news and lets you chat with a bot for more details about the news

Check out repo README first!!!

Link Github: News App

Tasks Status

Complete all the tasks

React Frontend

  • Render a full page application with three panels.
  • Navigation panel shows a list of channels.
  • Message list panel shows a list of message bodies for one channel.
  • Editor panel shows a text area input.
  • Editor panel is hidden if there is no channel selected.
  • Editor has a submit button.
  • Submit button is disabled if there is no text in message body.
  • Clicking a channel in navigation panel selects that channel.
  • Entering text in editor and clicking submit adds message to the currently selected channel.
  • Submitting editor clears input.
  • Switching channels clears input.
  • Channel list is loaded once on loading the application.
  • Initially no channel is selected.
  • Messages are loaded from remote on channel selection and updated to screen.
  • Messages are also stored to local state after loading from remote.
  • Editing is not required, only creation.
  • Upon submitting a message to a channel, that message is available for other users.
  • Submitting user sees message in message list after submitting.
  • Render created message in the list immediately before refreshing from backend.

ExpressJs Backend

  • Channel and message storage can be an in-memory database.
  • On server start, storage is populated with a fixed set of empty channels.
  • GET endpoint for querying channels.
  • GET endpoint for querying channel’s messages.
  • POST endpoint for submitting new messages to a channel.

Additional Features

Chatbot in order to analyze fundamental financial data of a company.

  • Channel Creation: User creates a channel named with a company's stock ticker (e.g., "IBM", "AAPL").
  • Data Retrieval: The backend system retrieves the company's financial data (income, balance sheet, earnings) from Alpha Vantage via API.
  • Data Storage: Financial data is stored in JSON files within a 'dummyData' folder.
  • User Interaction: The user asks a question by sending a message to the chatbot in the channel.
  • Data Filtering: For each user question, the application uses the LLM to query and filter relevant data from the JSON files.
  • Question Refinement: The application uses the LLM to reformulates the user's question to make it more suitable.
  • Answer Generation: The application uses the LLM to generate an answer to the user's question.

App Workflow

chat_app_workflow.png

Instructions

Backend

  • ENVIRONMENT VARIABLES:
    • PORT=4000
    • OPENAI_API_KEY=OPENAI_API_KEY
    • ALPHA_VANTAGE_API_KEY=ALPHA_VANTAGE_API_KEY
  • Setup:
    • Change directory to the backend directory: cd backend.
    • Create the .env file and place the variables into it.
    • Install dependencies: npm install.
    • Run in development mode: npm start.

Frontend

  • ENVIRONMENT VARIABLES:
    • REACT_APP_API_URL=http://localhost:4000
  • Setup:
    • Change directory to the backend directory: cd frontend.
    • Create the .env file and place the variables into it.
    • Install dependencies: npm install.
    • Run in development mode: npm start.

Run with Docker

Prerequisites

  • Docker
  • docker-compose

Run in Local with docker-compose