AniGenie

AniGenie is a lightweight anime recommender that suggests titles based on your watch history using semantic embeddings and fast similarity search.

Why?

Too many choices. AniGenie filters anime by matching metadata (title, genres, characters,summary, ratings ) to user preferences for quick, relevant suggestions.

How?

• Collect anime data via ani.py or APIs (e.g. Jikan).
• Encode metadata with SentenceTransformers.
• Use FAISS for nearest-neighbor search.
• Rank by hybrid score: 70% semantic similarity + 30% ratings.

Tech stack

• Python 3.x
• SentenceTransformers (all-mpnet-base-v2)
• FAISS, pandas, NumPy
• BeautifulSoup4 (optional scraping)

Example output

Recommendations for user 1:
- Fullmetal Alchemist: Brotherhood (Similarity: 0.85, Rating: 9.1)
- Steins;Gate (Similarity: 0.78, Rating: 9.1)
- Death Note (Similarity: 0.74, Rating: 8.6)

Contributing

PRs welcome: add data sources, tune scoring, or build a web UI.

---