Olivier/allocine-movies

🎬 AlloCiné Movies Dataset

This dataset contains metadata for over 42,000+ French and international movies scraped directly from AlloCiné.fr, the leading French cinema database.

It is designed for natural language processing (NLP), tabular classification, recommendation systems, and movie metadata analysis in the French language.

📊 Dataset Structure

The dataset is provided in a single CSV file containing the following fields:

Field	Type	Description
id	int	Unique AlloCiné Movie Identifier (extracted from the URL)
title	str	Movie title (mostly in French or original title used in France)
release_date	string	Release date in France (YYYY-MM-DD HH:MM:SS)
duration	int	Movie duration in minutes (e.g., 99)
genres	str	Comma-separated list of movie genres (e.g., "Aventure, Romance")
directors	str	Comma-separated list of movie directors
actors	str	Comma-separated list of main cast members
nationality	str	Country or countries of production
press_rating	float	Average rating given by the press (0.0 to 5.0 stars)
number_of_press_rating	float	Total count of press reviews contributing to the rating
spec_rating	float	Average rating given by the audience/spectators (0.0 to 5.0 stars)
number_of_spec_rating	float	Total count of spectator reviews contributing to the rating
summary	str	Movie synopsis/summary in French

Data Preview

id,title,release_date,duration,genres,directors,actors,nationality,press_rating,number_of_press_rating,spec_rating,number_of_spec_rating,summary
130687,Double Jeopardy,2023-02-03 00:00:00,99,Thriller,Lawrence Schiller,"Rachel Ward, Bruce Boxleitner, Sela Ward",U.S.A.,,,2.9,5.0,"Jack Hart vit une existence paisible..."

🚀 How to Use

You can load this dataset directly using the Hugging Face datasets library:

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("Olivier/allocine-movies")

# Access the data
df = dataset["train"].to_pandas()
print(df.head())

💡 Potential Use Cases

• Genre Classification: Predict movie genres (genres) based on the movie synopsis (summary).
• Rating Regression/Prediction: Predict audience (spec_rating) or press (press_rating) scores using metadata features like genres, duration, directors, and synopsis.
• NLP & Text Summarization: Train text-generation models to summarize or generate synopses from basic metadata.
• Recommendation Systems: Build collaborative or content-based filtering algorithms using nationalities, directors, actors, and ratings.

🛠️ Dataset Creation & Scraper

This dataset was scraped and verified using the open-source tool allocine-dataset-scraper.

The scraper features:

• Strictly typed Pydantic models for data validation.
• An automatic self-healing engine to detect and repair corrupted fields.
• Robust boundary checks to filter out abnormal or malformed data (such as negative durations or ratings out of bounds).
• If you want to update this dataset or scrape additional fields, visit the allocine-dataset-scraper GitHub repository.

📄 License

This dataset is distributed under the MIT License.