--- license: apache-2.0 language: - en pipeline_tag: sentence-similarity tags: - recommendation-system - hybrid-recommender - books - tfidf - cosine-similarity - lightfm - machine-learning datasets: - Amazon Books Reviews - Google Books API metrics: - precision@5 --- # Book Recommendation Engine ## Overview This model powers a hybrid book recommendation system combining Content-Based Filtering and Collaborative Filtering. The recommendation engine uses: - TF-IDF Vectorization - Cosine Similarity - LightFM - Hybrid Recommendation Strategy The goal is to recommend relevant books for both new and existing users. --- ## Model Details **Task** Book Recommendation **Pipeline** Sentence Similarity **Frameworks** - Scikit-learn - LightFM - NumPy - Pandas --- ## Dataset Training consists of - Amazon Books Reviews Dataset (3M+ interactions) - Curated dataset of 7000 books including title, author, categories and descriptions. --- ## Methodology 1. Text preprocessing 2. TF-IDF vectorization 3. Cosine similarity computation 4. LightFM collaborative filtering 5. Hybrid recommendation generation --- ## Evaluation | Metric | Score | |--------|------| | Precision@5 | 0.1688 | --- ## Usage ```python from sentence_transformers import SentenceTransformer model = SentenceTransformer("cs22/book-engine") ``` Generate embeddings and compute cosine similarity for book recommendations. --- ## Limitations - Cold-start users rely mainly on content-based filtering. - English language only. - Optimized for books. --- ## Future Improvements - Cross-encoder reranking - Semantic embeddings - User profile learning - LLM-assisted recommendations --- ## Citation If you use this model, please cite: Chokkapu Saketh, "Book Recommendation System: A Hybrid Machine Learning Approach", 2026.
#1 opened 21 days ago
by 
cs22