Learning-Resource Recommender

A lightweight recommender that helps you discover your next learning resource.

It blends patterns from similar users with content keywords, and explains each suggestion in the UI.

Hybrid implicit-feedback + content model:

• ALS (implicit) for collaborative signals
• LogisticMF (implicit) as a second model (replaces LightFM)
• Content TF-IDF (titles/genres) for explainability

Quickstart

python3 -m venv .venv && source .venv/bin/activate      # Windows: .venv\Scripts\activate
pip install -r requirements.txt

bash run.sh                    # quick metrics on the tiny sample dataset
streamlit run app.py           # browse top-N in a Streamlit UI

Notes

• Works with modern NumPy 2.x without downgrades.
• To add MovieLens or your own dataset, mirror the CSV schema in data/sample/.
• You can later reintroduce LightFM if you pin older NumPy/Scipy versions.