AI-Powered LMS – Data Science

1 | Project Overview

This project involves building a personalized course recommendation engine as part of an AI-powered Learning Management System (LMS). The recommendation engine uses sentence-embedding models to suggest courses to users based on their interests, previous courses, and learning goals.

2 | Embedding Models Evaluated

Model name	Mean Precision	Mean Recall	Mean F1-Score	Mean Average Precision (MAP)	ROC AUC
all-MiniLM-L6-v2	0.467	0.833	0.548	0.881	0.855
all-MPNet-base-v2	0.400	0.783	0.490	0.875	0.883
paraphrase-multilingual-MiniLM-L12-v2	0.400	0.767	1.481	0.871	0.848

3 | How the recommendation should work

1. User inputs their preferences: interest, past work experience, career goal, etc
2. Encode using an embedding model
3. Compute cosine similarity with course embeddings
4. Rank and return top-k results
5. Serve via fast API