PratikDhonde/letterboxd-all-movie-data

Letterboxd Film Dataset

This dataset contains a comprehensive collection of 847,209 films from the Letterboxd platform, including movie information, user reviews, and ratings.

Dataset Summary

• —Total Films: 847,209
• —File Size: ~1.12 GB (1,120,572,122 bytes)
• —Format: JSONL (JSON Lines)
• —Language: Primarily English, with some multilingual content

Data Structure

Each line contains a JSON object with the following fields:

{
  "url": "https://letterboxd.com/film/come-and-see/",
  "title": "Come and See",
  "year": "1985",
  "directors": ["Elem Klimov"],
  "genres": ["War", "Drama"],
  "cast": ["Aleksei Kravchenko", "Olga Mironova", ...],
  "synopsis": "The invasion of a village in Byelorussia by German forces...",
  "rating": "4.62 out of 5",
  "poster_url": "https://a.ltrbxd.com/resized/film-poster/...",
  "reviews": [
    {
      "username": "cameron fetter",
      "review_text": "as soon as this film ended i went online...",
      "likes": "11662"
    }
  ]
}

Data Fields

Reviews Object

Each review contains:

• —username: Reviewer's username
• —review_text: Review content
• —likes: Number of likes received

Usage

Loading Data with Python

import json

# Loading JSONL file
films = []
with open('full_dump.jsonl', 'r', encoding='utf-8') as f:
    for line in f:
        films.append(json.loads(line))

print(f"Loaded {len(films)} films")
print(f"First film: {films[0]['title']} ({films[0]['year']})")

Analysis with Pandas

import pandas as pd
import json

# Converting to DataFrame
data = []
with open('full_dump.jsonl', 'r', encoding='utf-8') as f:
    for line in f:
        data.append(json.loads(line))

df = pd.json_normalize(data)
print(df.head())

# Basic statistics
print(f"Most popular genres: {df['genres'].explode().value_counts().head()}")
print(f"Most productive years: {df['year'].value_counts().head()}")

Using Hugging Face Datasets

from datasets import load_dataset

# Loading dataset
dataset = load_dataset("pkchwy/letterboxd-all-movie-data")

# Exploring first example
print(dataset['train'][0])

# Filtering example - only films after 2020
recent_films = dataset['train'].filter(lambda x: int(x['year']) > 2020)

Use Cases

This dataset can be used for:

🎬 Movie Recommendation Systems

• —Collaborative filtering based on reviews and ratings
• —Content-based recommendation (genre, director, actor similarity)

📝 Natural Language Processing

• —Sentiment analysis on film reviews
• —Text classification (genre prediction from synopsis)
• —Review quality assessment

📊 Data Analysis & Visualization

• —Film industry trend analysis
• —Director and actor popularity analysis
• —Genre distribution and temporal changes

Data Quality

• —Missing Data: Some films may lack synopsis, cast, or review information
• —Language: Reviews are mostly in English, with some multilingual content
• —Time Range: Films from 1890s to present day
• —Scope: Popular films on Letterboxd platform (rating-based ranking)

Ethical Use and Limitations

• —This dataset consists of publicly available Letterboxd data
• —Contains no personal information (only usernames)
• —Check Letterboxd terms of service before commercial use
• —Possible sampling bias (popular films weighted)

Citation

If you use this dataset in your research or projects, please cite it as:

@dataset{letterboxd_film_dataset_2025,
  title={Letterboxd Film Dataset},
  author={Salih Mert Canseven},
  year={2025},
  publisher={Hugging Face},
  url={https://huggingface.co/datasets/pkchwy/letterboxd-all-movie-data}
}

Or in text format:

Salih Mert Canseven. (2025). Letterboxd Film Dataset. Hugging Face. https://huggingface.co/datasets/pkchwy/letterboxd-all-movie-data

License

MIT License - Free for educational and research purposes. Citation required for any use.

Contact

For questions about the dataset, please use GitHub Issues.

Source: Letterboxd.com