Buzz-AI

Codebase

Enhancing Georgia Tech Course Discovery Using Transformer-Based Semantic Search

• DataSource has all the data that we have worked with.

• processing has all the scripts for pre-processing and scraping data.

• src has all the code required for training and searching.

• Syllabus will have all the Syllabus PDFs

• User and UserData have code and dynamic files for Adpative Recommendations/Collaborative Filtering.

Setup

pip install faiss-cpu sentence-transformers numpy pandas pdfplumber

Search Queries

• -q: User Query - String (Mandatory)

• -l: Level - "grad", "undergrad" (Optional)

• -c: Courses - "MGT", "CS", "CSE" (Optional)

You can run:

cd src 
python3 search.py -q "Poker Theory" -c "CS,CSE" -l "grad"

You can change the flags as needed and required.

Adaptive Recommendations

cd User 
python3 main.py

Contributors

• Mehul Rastogi
• Akshat Karwa
• Pranay Begwani
• Vidushi Maheshwari