Food security 2023/2024 Chatbot

Architecture

• Worker : Worker does the heavy lifting. It retrives similar data from vectorDB (Chroma DB)

• Backend App : Its the app exposing backend API to frontend. It also updates results in redis cache

• Frontend : Frontend interacts with backend-api and gives back the result

• Rabbit MQ : Message queue has prompts and results channel through which backend and worker interact with for asynch communication.

• OpenAI API : We are using 'GPT-3.5-turbo' model and restricting the context with the information retreival done from vector DB

• Redis : To cache prompts and its result for faster response.

Demo

How to run

Prerequisites

Install following

1. Docker
2. Node.js
3. Python

Run RabbitMQ and Redis in Docker :

cd  docker

docker-compose up -d

To Run the Front End

 cd frontend
 npm install
 npm start

install pythin dependencies

 pip install -r requirements.txt

To run worker

 python source/worker.py

To run app

 python source/searchDB.py

open frontend in browser :

open in browser : http://localhost:3000/