Pulpnet_240221

IITK InfoBot

Pages Scraped:

Here are the sources used to gather the data from-

Institute Counselling Service (ICS)

• FAQ
• Events
• New UG Information
• New PG Information
• UG Information
• PG Information

Vox Populi – IITK 101 Series

• Games and Sports Council
• MnC Council
• Science and Technology Council
• Academics
• Academics and Career Council
• Campus Life
• Students' Gymkhana
• Course Overview

IIT Kanpur Official Academic Page

• Academics Overview

Sources scraped to test but not included in the final dataset due to lack of either relevant or extractable information-

• VOX IITK
• ICS Homepage
• CSE Department
• EE Faculty
• Academics at IIT Kanpur
• DOAA Homepage
• ICS FAQ
• IITK Main Website
• SPO – Companies FAQ
• SPO – Students FAQ
• PG Manual PDF
• UG Manual PDF

Methods employed:

• Web Scraping: requests, BeautifulSoup
• Text Embedding: sentence-transformers (all-MiniLM-L6-v2)
• Similarity Matching: Cosine similarity with top-k filtering
• Question Answering:
  • deepset/roberta-base-squad2 (extractive answers)
  • google/flan-t5-base (answer elaboration)
• Frontend Interface: Streamlit

Instructions on how to run the chatbot:

1. Clone this repository

   git clone https://github.com/pripky/Pulpnet_240221.git
   cd Pulpnet_240221

2. Create a virtual environment (recommended)

   python -m venv venv
   venv\Scripts\activate  # On Windows
   source venv/bin/activate  # On Mac/Linux

3. Install the required packages

   pip install -r requirements.txt

4. Run with streamlit

   streamlit run app.py

If you're running the chatbot for the first time, some models will be downloaded automatically. Depending on your internet speed, this may take a few minutes.