adi9235/conversational-chatbot-

Conversational-ChatBot

Problem Statement

Suppose we want to get the details of lets say 100 -200 pdfs so it will take us more time to manually read and get the desired answer of our question. So we want to construct a chatbot which takes our query and performs desired semantic search by using convolutional Buffer Window which allows bot to hold memory for previous chat history.

Solution

We will be going to construct a conversational chatbot that will take the inputs as number of pdfs and give a refined query based on the the users history. We will be using a vector database for our text embeddings and the database would be pinecone.

Architecture

Requirements

• streamlit
• streamlit_chat
• langchain
• sentence_transformers
• streamlit>=1.25.0,<2.0.0
• pillow<10,>=7.1.0

Future Scope

To built the UI for chatbot to work on Streamlit and also for chatbot to work on Textbase as I was getting the error of setting textbase. I tried both locally and also on Github codespaces but there was some error on vite.svg and that I was not able to configure. I was facing this error on Windows. I am aiming to work on solving this issue after this hackathon.

Issues Faced

Reference
https://blog.futuresmart.ai/building-an-interactive-chatbot-with-langchain-chatgpt-pinecone-and-streamlit

Thank You!