Research Assistant Tool

This project is designed to assist users in gathering information for their research. It consists of four main components: Scraper, PDF Converter, Chatbot, Speech Module, and a user-friendly Gradio interface to integrate all functionalities.

Components

1. Scraper:

Designed to help users gather information for their research. It performs the following steps:

• Takes a research topic as input from the user.
• Conducts a Google search to find relevant articles and information.
• Scrapes outlines from the retrieved URLs to provide an overview of the content.

2. PDF Converter:

Using "weasyprint" library, this modules enables converting a selected webpage to a PDF document, so users can easily archive and share it and chat about it.

3. Chatbot:

This the heart of the project, where the magic happens. This module is responsible for creating a Conversational Retrieval Chain using langchain:

• Instantiate an LLM model, either from local source (Ollama phi model) or from HuggingFace (OpenAssistant)
• Load given pdf file
• Split it to chunks
• Create and apply an embedding function on those chunks
• Create a vector database and fill it with the embedded vectors
• Create a retriever chain which is "history aware", which given chat history and a question, can reformulate it by adding context to it.
• Create a stuff documents chain, used to pass documents into a model
• Create the main chain, retrieval chain, which is built over the previous 2 chains

Then, it provides answer_query method, which gets the answer from the retrieval chain and saves chat history.

Chatbot Architecture Overview:

4. Speech Module

Provides voice interaction capabilities with the chatbot, allowing users to use voice commands to interact with their PDFs. This module includes:

• Audio to Text: Converts spoken input from the user into text using the audio_to_text method.
• Conversational Retrieval Chain: The chatbot processes the text input to retrieve relevant information from the loaded PDF.
• Text to Audio: Converts the chatbot's text response back into spoken output using the str_to_audio method.
• Voice Interaction: Facilitates continuous voice conversation with the user, providing a seamless and interactive research experience.

5. App:

All the project functionalities are integrated into one Gradio interface, to provide an intuitive and interactive experience for the users. The interface consists of 3 main tabs:

• Scrape Google Tab: Allows users to input a research topic and specify the number of links to scrape from Google. Then the user can take a look at the outlines of any of the scraped websites to choose the most relevant for his research.
• Convert To PDF Tab: Lets users input a webpage URL to convert it to PDF, or upload a pdf from their device. Then, they can load it to the chatbot.
• Chat with your PDF: Represents the retriever chain interface, which accepts questions input from user and displays chatbot answers, preserving the chat history (which can be cleared by the user)
• Talk with your PDF: Includes audio input and output components for user-bot conversation. Users can speak into the microphone, and the bot will respond with voice answers.

Getting Started

Installation

1. Clone the repository:

   git clone https://github.com/LaraKhansa/Research_Assistant.git
   

2. Install the necessary dependencies. Make sure you have python>=3.9.9 and <=3.13:

   pip install -r requirements.txt
   crawl4ai-setup

Running the App

To start the Gradio interface, run the following command:

  python app.py

Usage

1. Scrape Google Tab: Enter a research topic and specify the number of links to scrape. Review the outlines and choose the relevant articles.
2. Convert To PDF Tab: Enter a webpage URL to convert to PDF or upload a PDF file from your device.
3. Chat with your PDF Tab: Load the PDF into the chatbot, ask questions, and get answers while preserving the chat history.
4. Talk with PDF Tab: Speak into the microphone to interact with the chatbot using voice commands, and receive voice responses from the bot.