A content-based machine learning application that recommends the best relevant TED Talks based on user-input topics. Built using NLP techniques, a Streamlit web interface, and deployed via Docker for cross-platform accessibility.
This project leverages TF-IDF vectorization and cosine similarity to identify and recommend talks similar to the user's idea or interest.
- Content-Based Recommendation using TF-IDF and cosine similarity
- Text Preprocessing with tokenization, stopword and punctuation removal (NLTK)
- Interactive Web App built with Streamlit for real-time querying
- Dockerized Deployment for seamless local setup across platforms
- Data Visualizations using WordCloud and matplotlib for dataset insights
- Python, NLTK, scikit-learn, pandas
- Streamlit for frontend UI
- Docker for deployment
- matplotlib, WordCloud for exploratory analysis
docker pull hrishikeshuchake/ted-streamlitdocker run -p 8501:8501 hrishikeshuchake/ted-streamlitThen open your browser and visit: http://localhost:8501
This app uses a cleaned dataset of TED Talk transcripts and metadata from Kaggle, preprocessed for NLP. The input is vectorized using TF-IDF, and similar talks are identified using cosine similarity and Pearson correlation.
MIT License. Feel free to fork, adapt, and build on top of it!
Developed by Hrishikesh Uchake