BookMate

A personalised, mood‑aware and preference‑driven book recommendation app built entirely in Streamlit, using TF–IDF similarity, genre/author boosts, mood mapping, and personalized text‑based signals.

🌟 Features

Personalized Recommendations

• Enter favourite books, disliked ones, reviews, preferred genres, and authors.
• TF–IDF vectors + cosine similarity score how well each book matches your personal taste.
• Genre and author overlap boosts relevance.

Mood‑Based Recommendations

• Choose from predefined moods (Happy, Sad, Romantic, Adventurous, etc.).
• Or type a custom mood — the model converts it into a semantic vector.
• Uses both similarity scores and tag/genre keyword matching.

Elegant UI & Animations

• Custom full‑screen loader animation.
• Soft pastel theme, custom CSS, centered tabs, responsive card grid.
• Modern cards: genre badges, summaries, Google search button.
• Clean header with inline logo + About section.

🔸 Data Handling

• Automatically loads books.csv if present.
• Falls back to 5 curated sample books.
• Caches TF–IDF and dataset for fast re-runs.

TF–IDF Engine

• Combines title + author + genres + tags + summary into a single text corpus.
• Vectorizes it with TfidfVectorizer(max_features=2000, stop_words='english').
• Uses cosine similarity to compare user preferences/moods with each book.

Personalized Model

• Positive similarity boosts.
• Negative similarity penalties for disliked books.
• Extra scoring for genre matches and author matches.

Mood Model

• Maps moods → keyword lists.
• Computes similarity + tag matching for final ranking.

UI Highlights

• Fully custom CSS: cards, badges, buttons, sidebar, tabs.
• Full‑bleed gradient header with About button.
• Animated loading screen with bouncing dots.
• Responsive layout with adjustable columns.

🙌 Acknowledgements

Built with Streamlit, pandas, NumPy, scikit‑learn, and lots of pink pastel love.